SemEval2021_Research_Problem_Extraction.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# **Research problem extraction**\n",
        "This Colab notebook contains the implementation of solution to the task of `research problem extraction`."
      ],
      "metadata": {
        "id": "2oJZnirs5Ff7"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Problem statement\n",
        "We need to build a **research problem extraction system** using the [training](https://github.com/ncg-task/training-data) and [test](https://github.com/ncg-task/test-data) dataset from NCG challenge. \n",
        "The image below shows how a part of the training dataset looks like.\n",
        "For the task at hand, the relevant data is present in **research-problem.json**."
      ],
      "metadata": {
        "id": "2dZhsPXp6m-n"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "![Screenshot from 2022-07-02 10-13-37.png](data:image/png;base64,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)"
      ],
      "metadata": {
        "id": "DbE56aCR_bcc"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "An example content of the `research-problem.json` file looks like the following:\n"
      ],
      "metadata": {
        "id": "SiUtwhYIAEUy"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "![image.png](data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAABXIAAABbCAYAAADeKX3BAAAABHNCSVQICAgIfAhkiAAAIABJREFUeF7t3Qd0XNd5J/D/FAyAQRsMei8EWEESbBJJFatGiixbkp11lbPxJjluSewTJ7vrZL3Njk927ewmPkpckihKsl5ZXlmyZMuSrUqxSSwQwQaCIAkQBEGi10EZzMzb7wGDNhi8dwcYDAfUf454RM7cd8vv3te+uXOfZXttrQa+KEABClCAAhSgAAUoQAEKUIACFKAABShAAQpQIG4F7L19fXFbOVaMAhSgAAUoQAEKUIACFKAABShAAQpQgAIUoAAFAIvH41kwI9fpdNKGAhSgAAUoQAEKUIACFKAABShAAQpQgAIUoAAF4kTAGif1YDUoQAEKUIACFKAABShAAQpQgAIUoAAFKEABClBgEQEGcheB4dsUoAAFKEABClCAAhSgAAUoQAEKUIACFKAABeJFgIHceOkJ1oMCFKAABShAAQpQgAIUoAAFKEABClCAAhSgwCICyoFc7/kf4QsfqEa2MwFWaxLu/8F1LFhcd5FC+DYFKEABClCAAhSgAAUoQAEKUIACFKAABShAAQosXcCutqkHL3/jj/Bk76fwbN1/wC5XAhIzcmBR25ipKEABClCAAhSgAAUoQAEKUIACFKAABShAAQpQYBkCaoFcbQDt10aQuedDeGB9MRzLKJCbUoACFKAABShAAQpQgAIUoAAFKEABClCAAhSgQGQCaksraBrkP1htaskjqwJTU4ACFKAABShAAQpQgAIUoAAFKEABClCAAhSggJGAYmQ2GMi1KCY3KpGfUYACFKAABShAAQpQgAIUoAAFKEABClCAAhSgQEQCSpFZX/s+7GuwoKK6DGprMURUByamAAUoQAEKUIACFKAABShAAQpQgAIUoAAFKEABAwGTuKwXv/5iNR77+yuwbP+v2Pf5NVCK/BoUyI8oQAEKUIACFKAABShAAQpQgAIUoAAFKEABClAgMgGTuKwDd/z3fXjnp1/BujP/gO+/NRhZ7kxNAQpQgAIUoAAFKEABClCAAhSgAAUoQAEKUIACyxYwCeQCydnl2PzwH+LxbV14/dVTmFh2kcyAAhSgAAUoQAEKUIACFKAABShAAQpQgAIUoAAFIhGwNl3zQDPdwoFEBzA2OqaQ1jQzJqAABShAAQpQgAIUoAAFKEABClCAAhSgAAUoQIEIBKwP/uWRCGbZmod8IyibSSlAAQpQgAIUoAAFKEABClCAAhSgAAUoQAEKUEBBwH7xb+42T2axwSaLMExM+MzTMgUFKEABClCAAhSgAAUoQAEKUIACFKAABShAAQpEVcB0jdzJ0iyZqKrKwuCB5/DimTZcv96BvtFAVCvCzChAAQpQgAIUoAAFKEABClCAAhSgAAUoQAEKUCC8gFogF4m468+/hy+Xvo0v7KpAYWEZPvYvnVwvN7wp36UABShAAQpQgAIUoAAFKEABClCAAhSgAAUoEFUBi8fjWbDwrdPpjGohzIwCFKAABShAAQpQgAIUoAAFKEABClCAAhSgAAWWLqA4I3fpBXBLClCAAhSgAAUoQAEKUIACFKAABShAAQpQgAIUWJ4AA7nL8+PWFKAABShAAQpQgAIUoAAFKEABClCAAhSgAAVWXCB+ArnaBPrq6nHiSA/4GLUV73fjAlZjX2g+9J04hZNHezl+jHuXn1KAAhSgAAUoQAEKUIACFKAABShAAQqsQoHYBXK1EVx86ln86pVuBPxdqP/fz2L/kZE5D0ybQG/9eVxumfveKhS9Kaq8GvvCi973zqG52bP4Q/g8l3Dwm8/guf869eeFn1yBfyX7y3TMr2Th4fL2oe7rNUiwWGCZ/GNF1mdfwni4pHyPAhSgAAUoQAEKUIACFKAABShAAQpQIK4E7Prs19hEc62w2QGLzSoBJEz+3WqXv8cVBytzUwskF2Pb57LgC4yh7edvo2nFGxufY95e86d49ZU/xnqbBdZkFxJX3IEFUIACFKAABShAAQpQgAIUoAAFKEABCixXwP6TTg2fyI1BONWSgIREKxzOBKmzhoQkKxKS9b/zRYEYCVgdcOY4ZPg5kBSLoRevY96eiqz8fOTbYuTOYihAAQpQgAIUoAAFKEABClCAAhSgAAWWLWB/p2FCArkS3FrxlwVJ6U44XcmwWDU4M5zwpi4MIAda6vH6d45gZAxwZOWh/N7t2LDWOTtz13MFdf96Au09Y5gISGA4MwtFt9WiptYF+3R24/1oee0ELjT0YHgkAGtiIpKySrH1M7XIi2T64WgPmt9uwtWWXgwNjGJ8LACLIwnZd9+FvbemTdVJG0fP0VNoqLuGvp5xWFIzkX/rVmy5NRuOSOqjko9K21XqLKvIjl5qQsOhFnS0DWF8QmZIp6Qie+9u7N7tmrE27QuzMaN50fHGOzhzpg+ewXH4LRJILSxExd1bUFWeNNunKnUOjKDjYD3OHr+GwWEgMX8qn7Vr5owNqU+gvRGH/rYOvb0TsGfmoPhOGRtbMhBRzNKkL7RrjTj8fCN6+iZgdeWgoMKG/rMd8NgzUf7BPaipnm6b2pg3Y5z8XOp0/Zdv4thpoOjRu7BtXZLSZkxEAQpQgAIUoAAFKEABClCAAhSgAAUocHMI2KtKZY2DmLyscN12OzYnTJVX8MBtyElZuKiDJT0fa+8qRordKwHSepz76REkfeEDqHQFo6JJmSi5YydKM5Jgl+BWf/1JnPzFIdizH0RNiZ5fAD1vHcSJhlSse/AO5LvtEgMbxXCfFakRzsLUPJ1oOXIV2F6LLfekIlFmFAfGJMLsTgkGIf3ofWsfDh61ouyendhQkIDxlvM4/dp+HHM8gD3b9SCjSn1U8pGmmbZd4n2mddYwduYI9j3XBktlNdZ+sAZpqXYEhocxkTndrqkBYdoXpuPGh+HWDgw61+KWhwuR4Peg+/gZnP3RWxh//D7UlE2NBfM6T6Dr1X1457gVxffcgk05Ggbqz6Dh6X2T+Wwpn9OxATtcu7ZjbZYFw2fP4uzP3oYv8QHsWKf6ZYV5X2hD/ejpTUHV49uQ0XAER+oSsPbjdyJNXI+/fA5FlbVwT0aO1ca8KaOeIDCErouD8I4CnS3DCEggd+Heo5QTE1GAAhSgAAUoQAEKUIACFKAABShAAQqsQgH771TELhxkS01FchDJnpGGcCFkizsPJevzJ2dQZueMo+u79bjW7EXltuBUWlsqcjalzlC78v3oOn0IPW3ykLSSVAmcavB65PFNyQXIqshBZspUANhdsdTesSFjXQWKqsI4Dbfi3Dse5H7wIWzdEqxfUYbMCv05jp28irFt1Ui2KNRHKR+pv2nbp9toUGd/D86/1oqJih2495Nr4JxpVt4CINO+WLBF+Dcs6W7kVeZIn+Ygp9IF/MOraDpwBdWlFUicmZRtUOehKzh/fBjpdz6A7bvTJ4PouRXi3PMrNEo+a8sqMT0/1Vq8BhtvKZkcP3lSVqDrZZx9txUb11ZJX4Sv37x3FfpisqctSUgrdCN/wg17nRdpJdkokvadPD2EYZnhPBXI1bvMfMwr1EoyysLaR7bDdknaviuLQVwlNCaiAAUoQAEKUIACFKAABShAAQpQgAI3j4Cslhm/L4sEwZyJmvyUflzCs4mTATzN04WLb5zF5Uv9GBmVtXZT7PBJ4Mw5oT+2TX/ZkHfndhQ/cxwHv3sNWetKULKlAiVrJHCsEsiLgCPQ0YN+7wTGX3gRP3txzoaBALTMEUj1JHhoXh+1fFTabl55baAHvYMWuO8snhPENd8uXF+YbxUmhS0deRVONJ7qwUCgArkKax4EOsXZ70RpuR6oD76sEtAvT0HDyV4MBiSQGybODksKskpSoNX3Y0iGR7JKWQp9Om91jmCFNOlr/Sl++hcJk3+P+suCxLIqbCyLesbMkAIUoAAFKEABClCAAhSgAAUoQAEKUGAVCISbFBtH1ZbAmAToZgJjmgfNP92P03252HDfXuRlJ0AbuY6Gn5yCLHgw87Jml2Pn5wtQfaEVV07LrNlnGnG2aANu/VgNsp3RbJ4E7awpKP/o7agKfWCcTdaDDQYXzeujkI9i201bJ4FHPfZokaBjZK+Qvohs4zmp9QrInxUJdoZWKlhIRE1V6It5xYhLsEmhpfPfFKAABShAAQpQgAIUoAAFKEABClCAAhSIlkCcB3JDmunvR2+7H+l7N2Ptpoyp2Zn+cTilFXMDuZNb2RJlSYTqyT8buuUBWD88idMnyvGBvXNmdS5T0ZqbBZetBb3X/XBudBs/UMugPkr5RNJ2g3bpyxy40jRcOXsVIzJTeXZpBYONovmRBKR7Wj2w5LqQFm4WbZiypn26ZW1YrXhqaQVZ1BddLR5Y89xIXywf/xA6myVNfmgaK6wyO1eT2dQ+ifXa5gR6Vfpieu63XlVr1S588M+mKj33/TDNWOZbGsYvX8KlSxpydq1BdpgHBS6zAG5OAQpQgAIUoAAFKEABClCAAhSgAAUoEMcC9nOtA7JWaYZxEDJeGmBNh0tmvl491YBLeVXISrfJg8xkiQX/3Ar60X3oPXQk5CEnPwUOWwCjV/ow5rchRSK+EU3ONGt3WinW39KEAwcP4ODYOlRUuZBo82G8ZwQOWZM1b/IBbQr1UclHqe1mFZbPbdmovrsIV184jv1PD6Bqay7SUnTHUXjgRsX6YKBUISvVJIErF3H2iAQg3VZ45AFk564mofjjpWpr1uqFiM/aHRKMf/sQjts2oSQHGDx5Bo0dqSh/vARJOnNw8q022IOOS0lI8Emgt+4smjrTUPl46VSa6QpbEpCW44RWdxGNJ5ORl+iVtmdOtV2lL1QbHs10+trG0mdNvUCr14X7H8jmOrnR9GVeFKAABShAAQpQgAIUoAAFKEABClAgzgXse7/8Apqe+21kRTXCuUKttkpQ7iN74f31GVx4cR9OjQVgTXDAkSYP03IHJxdrPgR8XnQdO46LA14Jo9rgyMhAzh17ULN5+pFY0aqfHVn33oU7Mk6j4VgjTtSNwSdBwuTsXAmOSyBXL0apPgr5qLRdqVkWOLfuwV0p53HuUAsaf96EcVlj2JaSisyttShelw6HUj4RJLJNoPfd47jUL72RmYPSR29FzbpISrEj5/4PYHfSCZw5fASHR2S92Nx8rP1ELdaVJwQrYkdqqQSlT7Xg2I/Ow29NQEphITZ+eguqykMnnlvh3rsdazvfQ/PPD6A5IRmu7TtRIm1PsCj0RQRNj1pS6f+cNeloGZWHnclawYtNQo5aecyIAhSgAAUoQAEKUIACFKAABShAAQpQIK4ELB6PZ8FqpU5nVBeSjasGszIxFNBGcPGpl3A6ZTce/ljJ6pj1HUOe2BflQ93Xa3HbLz6GI8f+MzYrPPwt9nVkiRSgAAUoQAEKUIACFKAABShAAQpQgALhBEKnKoZLw/coQIGbSUCWnei5fh3XZXFga7ILuRmJN1Pr2BYKUIACFKAABShAAQpQgAIUoAAFKHBTCvAX2jdlt7JRFFhcwHf627i7uAAFBQXY8JXXML54Un5CAQpQgAIUoAAFKEABClCAAhSgAAUoECcClt7e7gVLK8RJ3VgNClBAQeCRRx7DCy88r5CSSShAAQpQgAIUoAAFKEABClCAAhSgAAVWqwBn5K7WnmO9KUABClCAAhSgAAUoQAEKUIACFKAABShAgfeNAAO575uuZkMpQAEKUIACFKAABShAAQpQgAIUoAAFKECB1SrAQO5q7TnWmwLLFdBG0PXOebS2Tiw3J25PAQpQgAIUoAAFKECBOBHwYfB0E87XD4BrCMZJl7AaFIiWAO9hoyXJfFaxwCoM5PpQ/63bkevOhnvyTw7W/MGr8x/YpI2h7dk3cPjtfmiBPjQ9+QZO1I/H/4l8YgyDzd0YGl18RPnbmnD0B2+j7thQ/Ldn8WYs/ESh7Qs3WuI7CmXdtM5zyQLD6DzWhp5e3801lpY4LOJqs5GrqH/iNbz5N1N/9v2yA4HlVDCexvwNOD57Lz6Lrz68C1WF+cjKKsRjT3XOjnnl+mjwtjbjzDMHceDv3sDb39+PI//vNNq5zPzUyFQYY8sZwlHfVrnfo15y+AzjrT7ha8l33xcCPNYtt5vfF9eQy0Va6e01CeSeu4JrV8eidI3L/cKwyxSuAW7a/UKh7YZ2kXyoUNZN6zzXifewkYya2KZ9P9/Dal34548Wyr1mPgqrduFDf/IsLnrD8Htfwx9W5QZjmRLPzL0N3zzhC5PQ+C27/sT6ROM0cfmpbcMf4Llnv4i1NgssSRnz22Cxwma3wGqdilNb5e8WPd10S8bacfIfGjBYthG7Hi5A4uQHAfS9cRAnh9Zg7yOFSLgBrdZ6rqDhpS5k/1YW0pJnajuvJv5BD0bHJmDtHZcap8F2A+o5WeR4F5qeb0LXwDgmJiywp6fCVVWMsp35SHVEXimVtkeea/gtVMqKG+fwTbix7/pH0X2kCS0NffB4/LL/JSElPwdld1UhOy38uL2xFVYo3duFxh83oGNwAn6JmNoSE+EsyEXR7koU5NoVMohykuRcrPtUhtTFi87X30PrMrOPqzFvdnxeZlsXbj6C1779Nfyo76N4at8fYXtGAhzpWbPnA9X6jHXi/MuXMFhQibUfciPR6oN3yAdHyiod8wuhlvWOyhhbVgHR3li136Nd7mL5mdUnltct450498w5dAaPh1aHA8m5WcjftQYlJYnBfUfDeH0d3nmrL+RLJgscm7dhzz1uWBWOq4ELp7D/pUW+qErIR83napBzQy50vLj+4mE0tgaQfttu1G5Lnj1mLNaHc96P33YpVD5ujnXBMbbfgY1f2HyDxoGCV5gk8XcNubzxHKaJ77+34mW/iOW5IIJeVrkGiJv9gvewEfTsTZaU97Ar36Hv53tYSzY+/o/1+E3vBPqPfxef/ezX8O17HsL3H3LOd3fciW8dPoOvaxr8DU/gsU+8vqR+sf/P9ybw9W03Imy5pPrObmRPkeh1LnLDXuDbYHNIcDFZAjAWDXb970mhwRgNE80X0Xw5G+vLV0/7HRtrsCt7FFb3DQzi6r3g92KkewzJu7ahplQutju7cfX4WdRdHsaWj1TBtRq/HZgzJOPGeZm7SfQ31zB0pB5n3gNy92xAVUESMObBYJtP9rFVHNAKTGB80IfUnduwdo0dPs8weo5fQOPzI8CntqIg1gFqSwKS3HJc0sbhSIiNa+zGvMrxOYojVxvE9Y4RZOx6APdUF2Lh90xq9dEGhuCZSELurnLkFsSmT6KowKwWCKj1+4LNVuwNlfrE6LrFP3U8TN+9E9XlVvhGpo6Hl34+Btvj21CUPmf827JQ9bFquGeuxfQv15Mw+TW6wnHVWlqNnY9XTKr6LjXixLtA+UfWIVu/htCPg2Gv8VasE2Yy1gav41qHCxW7A7h87jpGaisQyXc28douFTke61SUjNPE7nxqXI/pT5c7ntVKublTxdd+EaNzQZS7NG72C97DRrlnV0t2vIeNSU+9r+9hLUhy5UCiI8i990Hc4vpXNF8bkAkPzqnr4pkOcCAtJ0emZMq1b2cKQqOUqv1kP39BAhXbMlTTr5J0ViSmJSEpXWaOSCA3KT0JE6FX4BYHXMUJ6DrYiqKSNUgLd7OgTWDg5AW0nO7BYL8XlpR0ZG2tQtU21+yM3cA4eo83oVnSDA/LlOgEhwRg0pG7eyMqSoMB4pFOCQidR1f/OHwBKxIyMpCzsxprNgaDsfryD0/VoW1oahWn1mdeD87AsyChphZ7782CVZ+m/mQDev1TXWCt3ow7HsoLGRTy2WR9zqP5lNRHutaRk43CPdUoLZ2eRTOBXvkGoOXCMEaGvfD5ZfaM24383etQUZkU0YwTvSb2jHSkFwhegRvZZcmo/3ETLtTnY8ctqVN5mRmqtF0vyCyfSZYAxq+0oeV4O3quj8Drs8CWnAzX9k2o2ZYGi0pZ8ehsNn6k5b2HT5j3qc+DjkPn0XKuD2MTNiTlJSMgQ04/2Ci/NC8G22VglW3C2m25wRnh6XCVh+Zg0hd6cpWxeug0LjUNYXRIZspa7FLnHJklW4XiYsfsWFUaG6H1C/9vW1oqUnMk1JeTjowsPwb/uQmdlydQUBMM/6mU5R3CtYNNuHJhQGbPaxKLkBmgmflY+5gEO6ajiKZtD1+/ee9qw2h5+giu5tdOzX7TP/R34sz3TsN7+x7U1soMsrgc8wrHZ4XmKyeRbzvlP1hti60kZFKfEMORn7yOK5OFW5Fxxx5s2x6cqTc+gPZ3r6Dz6iBGhuSXCuMytVvOB67dO7Cl1jk1Xk36XetsxalfX8bAgA/WdBeyS2wYbOrFmD0NhXdvRmX5nHFvBqAyDk3Hs8L5QmWM6XWNRlmTbVY4tpiWNdV/ptcJZsYzn3vR82YdGho15P7GDqytXPh1gXlWJuNQzyCW1y1SnC01FSnZ+sVRGtJTxtDzdCuG+mRnmhvIlcvPpCxJF+4aKthow+OqQ37RkTWV0Nc5lUmiW/KL6MQ0VzcafaFh5Nw1eArLUbMxgIFjLejoLEdlXgRf4Ki2S+W4YXoNoLCf6kRmxwTVY53K/qXSrrndtpy/q9TH1FAqYFpnBWela0iFfHQPlfsLJTez8Sz1Mb3WUkmjYjjdLqP7FEmj0l8q5wLJSmu7gCN/fxbj8hPUhEw3CvauQ3lFBPc7qvuFyfl9sqtMx5hSh8b2XGDWFyrXAEr7hcrYUNx3FBl5DxuDWIHZ+IFin/Ie1nRUG15r6VurnCvNrhP0fFSOdWa1fV/cw8oEDbn11PSb0BV62f/4oZstiKtLWZC2Ywuq7FM3BVl3bEGmM/QC3IKU2iqkvC2BorOF2LI5dAqpBHHeqUP9KSsK9qxHufy8eqKtFRcO1qPBcSs2b9IvAvzoP1iHUycl7rN7PSry5OZttAvNr7ShX193VAK5k6UmpSL3lvXIS3PAJjvRcMNFNL1+CrbM3VhTID1szUDZR/egsPMyTv6qF5kPbEVptr6lzGyRn3hPhh+S87D+0y74JZDW8epiP7X2of+A1Oe0Bbl7JZDsBjwNl9D8Yh0mHpX1IYsl3q/5MXK1F4OOYmz8zWwJSI9LsPoiWl45A8entqPYFeqkPvIsrkIUll5Ew8UujOySGzGLgqFK28XZvC9kVnCTzAh+pROW0hKU3r0GTqcN2sgIfBnBQItKWfHobDZ+pIvM+3QC3W++J33jkPEsP1XNkAv7K1fR2q7ev5Mp5Vs2Z5YD2oXr6OiVLwlmp2HNyUihL6A4Vttl+YakEmy8R8aqfwz9py7h0gt18Mp4XlOk798qYyPCNgaTWxz2yUC1X9Za0A/B+v6uMg4H3j2JxgtOlH2gFlkuGYNy9zA6YEXyzMR/hbYvrcoLt4rLMa9yfF7YlKW/Ewzkyk/Xw79M6mNgaHXO3gxqI324drIL2LQWVXuS4ZBfgmjjsjCSazqNeb9rniEM9Cej+NG1SLtwFqfP2FD2cC2ccmw7t+8ysmX2YsZizZjXOA3m41BhPKucLwx8Zs5fKvuOSlmyJ5oe51XKmrQy6ffwgyX8u4FR9F32SPAe6GkbgSaB3MjPpCr1ieF1y0xL5Wdf+i8UTnZiLDVXgplKAzC8k66+4Li6aNKlfRCNvtDX3msag2unGwlODdmFjWi9MCTXeOlL6FfjZpgfN2R7s2sAxX3H9JhgsC/PHusUjhtSZaV2GdMofqpWH1NDlTqrOKtcQ6rko3p/oaJkNp71+phda6mkUTFUufbT22Q25pXOBUGcFDfK9uTK7P7g5JxXzsLx6ZBfFhg5Ku0X5ud3vYjo7RcxPBeY9YWBD+9h9V5XOUYpXNusVmez8aN0POQ9rNEhKtxnC6+11Mah6XWC6jE8XKUifS8ux7zKtXqwoRZ9mdcVDuQOjkqIInS2aqTQcZje6nTOrJtrTwtZl2K6vo5MlO3KwNEjl9G/fu38Vox0oOXEGNx370H1+uDsmrxUBDoOoOFcF7wbS5A42oHWU6Nw7rwFG3YEZ6DKTxKvWdowb11jqxOZ1bN1SMuR9XjPn8LANXkAW4EeZLTCkeFEwqhd/iazcNNS4MwMuQ2U2YiJmXog1i4/tV4E3NOBy6dHkXrLrVhfmzJ5w+EuTkGg711cPtaB0uKimZ8VW9IykF0ua9hJGle2BKSfOi83oBMSyF3KTKLp+lil3kny7fco9MloKeMKhhJcMW27Sl8EBnDlYAd8Jeux68NFmP2Vv0SzZ14KZcWjs8n4mW6eUZ9qQ9fR1uhFxm0yW2zrVGDbXWLBQH3//LG6yNCaa5i5dwuqx87i4tOHcK2yAIWbi5BbnDy7XrNfoS8iGKtIk5nwJZlTY7UkFXjmKFplPJcUyc/kVcZG5BEVBEaH0f1uK/rtLlSWBWezK5UlP3cbmZCbECdcJS6kB9e6ziiZAxtJ2037wyxBfI55peOzWdMUP/ddP4SDjRaUPViy6E9XjOujYDhTFytSZJ/ILQsT7FLo98mf1sisyxQJGLl96bCfmZC1ml3IQSYuNMr66DK0MkK/cwzroDAOVcZzMG+jY4s+Mzkqx3CVslSOLSrtCh4TjPs9LGz4N63pKL1/HWytQObWjCUH+5TqE6vrFpnt1v3rt/DWr6e+CEFCKooerIQ7dKasvwOnn+iYdbHlYtPntiA3zHVK2ONqeNGlvxuFvtB6OtE15JJfMumN0OAuT0PTiU4M7ZFfIIXZtZde2ektDY4behKTa4DpU5zxfqpwTIjWvjxzzjVp1/Lh5FtsletMdUP9eLboMTxYX0NnlWtIlXykXUr3FwqGyuPZ6FpruhyVNEaGCufBybsQszGvci6YdpZZuHmV8stG+bfLPYE++aVVz5UJFG1Svd9ROMeptmuyTuZjbJrb8P+xOheY9YXKcUNlv4jA0HAfNEQz+pD3sCsSKzAdP1N9YtSnvIc1GrcLPwt7raV0rlS4TohgP11Ys0jfUTj23oBji9K1ut5UW5ks12jDCwcP4fpn/w0Kl7p+ggGb/Q/+cj+OfufOyH5ebZDlbwSDAAAWLklEQVThavsocb08zKiuDpcbSlE6p/KB7gEMTfgw8doB7Htt7gcyOy9jDOMS/07Q0/gTkVc6FTRdrO3aSD+uHr6Ea63DGJMZO3aZKer3yZIPvmU9h35BcYGeQQz7k5BfHPwZr55CP4AWJ6H53CA8fgnkhrkJscjyA0lyPTMsUQJNQr1LiHvNqcvs9HEVw6kHzS1oyrw3VPJxDA9gcNgi62DKt+7La4BxZeTTWDsvZfyE9qnWO4xhLQkFsqbtsnkc6XJDfyvy+vvQdfYq2n75DppzymWGd4U++RCaQl8s1VB+6wt3SRJaGvXxXAi7wn6qMsamOl1/4KHs729K4MIvSyKkZclsyE0oCs5SVxmHifLFhPuWdcj7xTnUP9WNjMo85K4vlGOEE/LMxcmXUtsNfqZsOkBXIIFSnVf82BJpw7x4809uweP/chWWLf8ev/hs+eSN3I16qRjO+33M9I46NR18qtrKv85RGIdL3HdCjy0qxxO1fWdhz4SWBZVjyxLbtbD0SN6RJYqKilFRFMk2S08bi+sWPdiQuXcHqius8MuFi6e5Fc0vH4P/N3ZiffX0Uk3SBpsba35rzhq5+hI48y5WjY+rS1dYbMvl9oUGz4UujOeXwJ2slyH5lWYjZX87urvXID1XZcQvVrelvR+NawB9woDZuUmldkvdl1XyXkoa1fosxVClPqHHqKWOjtB8NMX7C/M6LnE8h1xrhbt3CL0ecyhct6icB/V8zPpL5ToznI3FmTz566hh+cJ9+fc7syWotitcnZbzXizOBWZ9sZz6z91WyTDMRVzovrPUfXCqLryHjXasYCnjJ7RPeQ+rspcZX2upnSvNrxOU9lOFc4FKi6KVRqnOK3FssZbgM9/8Cp556EuoLfsmPvOvR/FX96h+gajWenvnyVNo99+JyjhDV6t+FFLZ0lC8w4Wjx67I08hD8rNI4OuBWpRkh7xvlQdw6B2ur70oF/kyc3rxlzaK9ldO4OJAJspv3wx3plxBjPWg+aWLYWZCGmUULEIhSfjKGG842QblIEH4EvS1C0f6JFKdlixPdJ+ur4nhTFbG9ZOnpxj3xXR40iSbqeIUEikkCa9gvGHEzhGNn/k1mleWVGuyi5fdx9NlyAMEXfpaY7K+8uZrOPNj+en3MTduvV0PRQUNjCnC802+a7JhsCEzTTEdGwZFzftI1jy9dQfWVdvgqT+NhksyM1IWtZ1XG4WyrJkF2PDpLJS0dKDjvMyS/8VltORVYNPDlZOBbuOXSdvnbTx17NH0hY5NXwr5KiQJX4zxhhGP+fCFLOFdB/Z87UW8eu/f40u//3/wTwc+j7+5T19WPh5f4QyD55ZwHyk0QWkcmo3nRYZW+D41qWhUygqWYVKU+flCATCek6z0dUuw7ZNr5GbpF4eyRm6RzDTuO4BGWWKhskp+kTTTBwlIzklD6qLXkArH1Xiy1kbQ0+KBr6sRh59oDNZM/3JPzp8tI6jINf7iPupNidY1gFRM6Zig0gCzfVklj2imMavPMgxVqhn2eGh2jAqT8bx8VO4vwuSx4K3ljGe9DZP1MHippDHYfP5HQTSl/lI8FywoW86rk/dvCz5YwTeWMBhUa7PS5wKlvpiurEI7FZKEb7rxhmH3wfAZGbzLe1iz+6+InSMaP/O7hvewwViTwYid/5HCtZbZuVIyXN51gvF+GtLDk/Gzm/oeVhvAm99/Eg1Vn8NTT/w+bl0T3SCu7mlPzM+HO0wUWnncrPqEFiSuLUf+sXq0X5YZsjlTDbJmpcuNyTUMdvuRVJ0ediaX5pIlECzXMdgxBi0/uA5rqIesTTXQ6UfK9iqUVgdvAORpmfqTmOctvzDZG1Ypxw+fV7/CWGxnsMKmz3YZ98kqJfIwsznlTde5X1+jLz9YVmAEfW1jsGanTT2MZAUvXrTBazLrOICUbTniIi1QMJypvkHbVfoCqWlIS9HQITNpxjYUGs/KNShrljNOnCMZP3PGQuhfpw37ZFZ4oCAt7HgO3Ub135Y0NzKzLOjpG5VQfoZM0jDviyWPVbkoGGgfk7El41mOW0pjQ7Uhks6akizLgzjgvH0jhtuP48KhPLju09eSjrAsqwOplSWTfyr6WnHy6QuyFne+PBhLnloZ3C8M99PpOstZTl9fB16/HBn0OXJzXrJesUNmjPn7RzChZc0JrIRp8Goa82Gqv9S3krJKsfGB38PHtjyJH755FhP33Tr7oMqlZrrE7SLqd72vyzbgti9OFbbk324ojEOjc1xETTUYY9HaTy0RHFui1i4lBA3eq1dxtVVfWqEIrgVr8itlEkGiFb5uCVsTCejoA1G/8l7s8iTsdsbH1UU2Wcbby+sLbaAb3b0pKPnwZhTOTJHXMHz8BBpaujG2KwXBFXOWUccINo3SNcBMiQbHBJVujda+HIGAYVKl+kTb0LBG0x8ufg2psrlF5f5CISOl8Rwun5BrrXBJoJImZEOl86Df/L5J5VwQts4r9KZSu1ak7BU+F0Sy7xhcA8w2ffH9QsmQ97CrK1YQyfgx2D+mxwbvYQ2Q5CPewwpCPMXHfOew/3Afan7vd/HgptKoxl6mR4L9k7/7G8hQuXozHjur+1N7Boq2pKJ9X99MIBcp+Sjf2oZ6CfDWj5ehsCwVDpsf3r4x2CuLkZUm3+qm56Oo4jIaDp9Ck6UC2ekaxtqvYlBudmYm3llTkJZlRVdjC65mFyEjTX/40RDG9KhMyMuSLsEphxddx5rh0jLhkKcCenxpKKoOrr87md6OlJxkaHVtaD7tQLbTh7GAPNG8StJInctqWlF/5BTO2SqR55abj3OXcLk7GYWP5s0L+oaWvdR/+wYGMdRhgbenB+1H5Wnr6SXyhPZgfRUMp8s1bLtKPjYXSnbnoPPVczjxogfFGzKRkmyF5pUHTSEdhWtmZ9EYljUDESfOEYwfwz7Ux8ZmGRtHT+KMVoHComRYvQMYk40i+h5HG0fX4WZ40rOQkZUoywX4MNrWhivtsvj33uAXHip9EcFY1WSfulQvARJZ4mD0QjNaOhzI/WDeVOBSZWwYwizyoT0dJbcX4PqLTWjdJD8d1h9KqFRWAP11jei1ZyEzJ0m+mwlg/PqgrBktDzuT3xtPHmojaLss4CIPl0uEdroNLQ0OZCVOTI3nSn08JyCz0gXLvhY0HklAcbH83HlsQB5fuDDOsqrG/JwuGT36bXzk8e/h+p6/wPNPfhLlEQ3W6Ywc8uAx+SGE/Dx8Be8BFhlIc95W6PclB2wXlK42Ds3OcQuyNXjDcIwp7TsGmU9/pHhsiWa7FGolP0YZROurjbgyAHRMpOLWO2W/VNpwGYlW8rolWC2/R9Zl7peloMbGMNTUgkutVrjuzgq5lpBrkJ5heObOyJUvmZLcibPrpk83M9xxdRkEYTddVl9IELhVnuLtyEBFiXxRP6dNjrIMWBt70OspRVHqivfubNOidQ0gX7OanpvCgoa8Ga19WaWsmTQGY0ylPlEzjKTSBteQCtko3V+Y5qM4nlOmMjK81gqePA3TmNZHEiicByUaYX7fpHIuUKlPtNKotCtaZYXms5LnApW+CNbH8Bpgps4G+8UNMOQ97ArHCiIYP6HDet6/9bHBe1hDonkfhrvWUjlXqlwnRLSfvr/vYSUIhXEJtCTqN6Er9LJ/4wNpK3/DsUKVj162FiRvKEH20X7IM8eDLxsy9m5HbdpFNJ+6jPOnvfBPLqicKQ9ZKtZ/aSjRElkf975a+Pafx5VD+hIVNiTnTi6qNjVrZfL/ThQ9UAPfgWa0vf4eLo7JYgwJdiSkpMEtT7Sf93Jko/LeEkwcvIrGF1ugOZKQXrMe+RLInV1yTgJm29ajvOsc2vadxHVbEtIkTZ4Ecu1y2+S6fTs2J53HpbozODUqoZ7sLJR+qBplxVFeYdkmsxazktD13ns4flTKkSC0q2oDtu8sQOrMw3gUDKcBDNuuko8FSRtqsMN5BS3H29H6+hV45cFAVlkPK33DWuRKIHfm2SuGZU1XKE6cIxk/80dTyL9kbNy2HVucF+QLgCacPuqDJlO7EzNdyHMFA4yG209/GJAJWfJ03vpGtPWPw6dZpe/lwUx7tqJy2/TazCp9EcFYtfkweOIcrg76YctwIf/+GlRWTvemythQatiCRPaScpQWXcPFw1dR9FiJzPJWKcsPzSc+p2T/HJyQU6IV9vRUZO6qwZrphyZGtJ9akL59PUq7G9H+uvxqwB7c3yWQq/daUs0m1Iycw8VTDTj5rvRNgszSlT7NDe3T1TTmZ3oigI5jb+G97kF4f/UCDg18AuWhD4Fc0GtGb9zQMK5ULIIxb9QMpc/UxqHpOU6prGAiwzGmsu+oFKZ2bIlqu1SqZU1GZlkKrjXKg7HmrlGvsu2S06zgdYstAYmpVnQcPIJ3DsivAuQJq4luOVfcux3lG0N+feTvwYWne+a3IrEAm39/E0JXpdITLTyuLhkg/IbL6gsf+i5LND43H/Kd+7yXLTcDKdpF9F7xoWhDmCe5ha/N8t+N2jWAyjFBpbrR2pdVygqmMRpjNoX6RM0wgjrL+Xnxa3WFfFTuL0yzURzP64MZGV5rRZDGsF4K50Gl/lI5FxhWJMofKrQryiXOZreC5wKlvgjWxPAaYLq2RvtFDA15DxubWEEk48dw/+A9rCFPmA8XXmspnCvlt5+8hw2DGcdvWXp7u2/03W2EPD7Uf+suPPirR/D6G3+KjYuuyxZhtlFKrg1dwYl/vgDHA3dik6y1yRcFVlrgkUcewwsvPL/SxcQ+f20MbT89hIvOTbjjobzIZg7HvrYscQUEtKFT+Nvf/ii+0fZJvHTov2HnUuInWiee+sg2/EXJv+DMd+9dkV8mrEDTmSUFYibA65aYUbMgCixZYMX2U5VrLZU0S24ZN4wXgRUbY/HSQNaDArES4DEzVtLxW473DXx582fQ+h/r8Nxn8xadOOs7/T9w1/0v4sGX9+E/1UY28TKy1PFE5fOgt7MTnTZZ4iApAznpKzdtefFmBzDU2IZhm6yZliKUYx70nGjBQGIONhUziLu4Gz+hAAUoYCzgvfQqfviDH+Af38vEJ773JexYShBXL8LiQkWlG0P7f4FXzm3AbrfMWpalQVyTT6zkiwLvNwFet7zfepztXY0C3E9XY6+trjpzjK2u/mJtKUCB+BeQZVb7uzEoPwvve/clvDOQidsrM8MEcb0Y6hrAqDxQ1N8jD9hdYsNWbSDX3/AEPrzpCWm2BZmf+r84+8T9mPlF/xIxIt5M1r4YbZOn0l8clvUX5WfNjkQ4C3Kx7tE1kGVs+aIABShAgSUKWIdk7d/Rnfjaz36I36p1hzkJqmbswO1f/Q4+1/xf8NV7tqN33Io7/+oEnvud3GXkqVo201EgzgR43RJnHcLqUCCMAPfTMCh8K6oCHGNR5WRmFKAABaB145nf3Yo/fssvS5JVYM/n/wpfvT3MZFPv2/izPZ/Cj3qDT0exr1sSnsXj8SxYWsHpdC4pM25EAQpQgAIUoAAFKEABClCAAhSgAAUoQAEKUIAC0Rfgb0ujb8ocKUABClCAAhSgAAUoQAEKUIACFKAABShAAQpEVYCB3KhyMjMKUIACFKAABShAAQpQgAIUoAAFKEABClCAAtEXYCA3+qbMkQIUoAAFKEABClCAAhSgAAUoQAEKUIACFKBAVAViF8j1/hL/LtsGi8Uy9SdhE/78+FKf0RZVA2ZGAQpQgAIUoAAFKEABClCAAhSgAAUoQAEKUCCuBayN/hjVz3Ef/vpsO65du4a2176KDbYYlctiKEABClCAAhSgAAUoQAEKUIACFKAABShAAQqscgF77OrvQHpuHtKlQF9HCuyW2JXMkihAAQpQgAIUoAAFKEABClCAAhSgAAUoQAEKrGaB2C2tsJqVWHcKUIACFKAABShAAQpQgAIUoAAFKEABClCAAjdQgIHcG4jPoilAAQpQgAIUoAAFKEABClCAAhSgAAUoQAEKqAgwkKuixDQUoAAFKEABClCAAhSgAAUoQAEKUIACFKAABW6gAAO5NxCfRVOAAhSgAAUoQAEKUIACFKAABShAAQpQgAIUUBGw7i58FP90XVNJyzQUoAAFKEABClCAAhSgAAUoQAEKUIACFKAABShwAwTsP377CezOsdyAolkkBShAAQpQgAIUoAAFKEABClCAAhSgAAUoQAEKqAhYy6uKkWFTSco0FKAABShAAQpQgAIUoAAFKEABClCAAhSgAAUocCMEuEbujVBnmRSgAAUoQAEKUIACFKAABShAAQpQgAIUoAAFIhCwruNs3Ai4mJQCFKAABShAAQpQgAIUoAAFKEABClCAAhSgQOwF7LEr0ovBzj6MBDT4uz3w8flqsaNnSRSgAAUoQAEKUIACFKAABShAAQpQgAIUoMCqFohdINf7Gr6y8UP4p57AFJh9Ix5b1XSsPAUoQAEKUIACFKAABShAAQpQgAIUoAAFKECB2AhYPB7PgrmxTqczNqWzFApQgAIUoAAFKEABClCAAhSgAAUoQAEKUIACFDAV4MPOTImYgAIUoAAFKEABClCAAhSgAAUoQAEKUIACFKDAjRVgIPfG+rN0ClCAAhSgAAUoQAEKUIACFKAABShAAQpQgAKmAgzkmhIxAQUoQAEKUIACFKAABShAAQpQgAIUoAAFKECBGytgHMjVOvDDB5JgtSbAmV2Nu774I5z33tgKs3QKUIACFKAABShAAQpQgAIUoAAFKEABClCAAu83AevHnxlevM2WXHzmx5fR3t6M408+iJ5//CN842XP4un5CQUoQAEKUIACFKAABShAAQpQgAIUoAAFKEABCkRdwPqd33QaZGpBcmYe8vOLseHBD2OvewTtV/sRMNiCH1GAAhSgAAUoQAEKUIACFKAABShAAQpQgAIUoEB0Bawl6carK8wWZ4NNkmqaFt0aMDcKUIACFKAABShAAQpQgAIUoAAFKEABClCAAhQwFFCN4gIWi6yVy0CuoSY/pAAFKEABClCAAhSgAAUoQAEKUIACFKAABSiwAgLWMdVMbZXYUG1Dw759aPepbsR0FKAABShAAQpQgAIUoAAFKEABClCAAhSgAAUosFwB65+9ZfCws7m5W8vwe//rayh/5d+iIqMUX/iVd7llc3sKUIACFKAABShAAQpQgAIUoAAFKEABClCAAhRQELD+7LmjUArJav349V//HU6v/TKePfw2vnGnQyF7JqEABShAAQpQgAIUoAAFKEABClCAAhSgAAUoQIHlCtiVM/CdwZv7e7H1S1/Ch7aUQ31xXeUSmJACFKAABShAAQpQgAIUoAAFKEABClCAAhSgAAXCCFgf/cguKM2t1cYxJgvqJiYmhsmGb1GAAhSgAAUoQAEKUIACFKAABShAAQpQgAIUoMBKCVi/dVfqSuXNfClAAQpQgAIUoAAFKEABClCAAhSgAAUoQAEKUCAKAtYk5UwmMOEDrDYbLMrbMCEFKEABClCAAhSgAAUoQAEKUIACFKAABShAAQosV8BkjVwNo32dGBifQO+h53Gg3427qtwM5C5XndtTgAIUoAAFKEABClCAAhSgAAUoQAEKUIACFIhA4P8DvRGlDR88h2kAAAAASUVORK5CYII=)"
      ],
      "metadata": {
        "id": "7SauaYFiASA1"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Now our task can be roughly divided into following steps:\n",
        "\n",
        "\n",
        "1.   Extract the relevant test and train data from the dataset.\n",
        "2.   Preprocess the extracted text data.\n",
        "3.   Build a Model.\n",
        "4.   Train the model.\n",
        "5.   Evaluate the performance using F1, recall, precision score and additionally Jaccard similarity index.\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "eykazIinAgyt"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import csv\n",
        "import os\n",
        "import json\n",
        "import pandas as pd\n",
        "import re\n",
        "import string\n",
        "from sklearn.model_selection import train_test_split\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "from tqdm import tqdm\n",
        "from tensorflow.keras.preprocessing.text import Tokenizer\n",
        "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
        "from tensorflow.keras import *\n",
        "from tensorflow.keras.layers import *\n",
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "import random as rn\n",
        "from keras.utils.vis_utils import plot_model\n",
        "from keras.optimizers import *\n",
        "from keras.callbacks import ModelCheckpoint, EarlyStopping, TensorBoard, ReduceLROnPlateau\n",
        "from sklearn.metrics import classification_report\n"
      ],
      "metadata": {
        "id": "xdB6b-zihyaQ"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# 1. Get the datasets and extract the relevant data from it.\n",
        "In this section, we clone the test and train dataset then we extract the research problems.\n",
        "\n"
      ],
      "metadata": {
        "id": "CM1r5sgH1fXH"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "cd /content/drive/MyDrive/ColabNotebooks/"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iGQ77Gwfzu6S",
        "outputId": "60e9612b-9dc8-4c98-e96e-07a0dfdff046"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "/content/drive/MyDrive/ColabNotebooks\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "I9FpnS-EX8Lw"
      },
      "outputs": [],
      "source": [
        "!git clone https://github.com/ncg-task/training-data.git\n",
        "!git clone https://github.com/ncg-task/test-data.git"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def save_csv(input_name,output_name):\n",
        "  with open(output_name, mode='w', newline='') as rps:\n",
        "    writer = csv.writer(rps)\n",
        "    writer.writerow([\"research_problems\", \"sentences\"])\n",
        "    writer.writerows(input_name)"
      ],
      "metadata": {
        "id": "qJZ3IO43hMnm"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import os\n",
        "import json\n",
        "import csv\n",
        "\n",
        "def save_data(root, output_file):\n",
        "    research_problems_data = []\n",
        "    for path, subdirs, files in os.walk(root):\n",
        "        for name in files:\n",
        "            path_add = name.split('/')\n",
        "            if name == 'research-problem.json':\n",
        "                file_address = os.path.join(path, name)\n",
        "                with open(file_address, 'r') as json_file:\n",
        "                    rp_data = json.load(json_file)[\"has research problem\"]\n",
        "                    for elements in rp_data:\n",
        "                        if isinstance(elements, list):\n",
        "                            research_problems = elements[:-1]\n",
        "                            from_sentence = elements[-1][\"from sentence\"]\n",
        "                            for rp in research_problems:\n",
        "                                if isinstance(rp, list):\n",
        "                                    rp_prob_sent = rp[0]\n",
        "                                else:\n",
        "                                    rp_prob_sent = rp\n",
        "                                research_problems_data.append([rp_prob_sent, from_sentence])\n",
        "                        elif isinstance(elements, str):\n",
        "                            from_sentence = rp_data[-1][\"from sentence\"]\n",
        "                            research_problems = rp_data[0]\n",
        "                            research_problems_data.append([research_problems, from_sentence])\n",
        "    save_csv(research_problems_data, output_file)"
      ],
      "metadata": {
        "id": "iR685l85ZY7u"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "save_data('/content/drive/MyDrive/ColabNotebooks/training-data', \"train.csv\")\n",
        "save_data('/content/drive/MyDrive/ColabNotebooks/test-data',\"test.csv\")\n"
      ],
      "metadata": {
        "id": "R6t9pxr4ff_6"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_data = pd.read_csv(\"train.csv\")\n",
        "test_data = pd.read_csv(\"test.csv\")"
      ],
      "metadata": {
        "id": "3SEU8-2mjRnk"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_data.head(10)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 388
        },
        "id": "lhk0YhwYkLfd",
        "outputId": "babde172-ed48-4d69-d08c-56cfdd019674"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                                   research_problems  \\\n",
              "0                                 Text Comprehension   \n",
              "1                                    machine reading   \n",
              "2            Large - scale Simple Question Answering   \n",
              "3                   large - scale question answering   \n",
              "4                          simple question answering   \n",
              "5                          Simple Question Answering   \n",
              "6                    Task - Specific Tree Structures   \n",
              "7  task - specific tree structures only from plai...   \n",
              "8                                 question answering   \n",
              "9                          Question answering ( QA )   \n",
              "\n",
              "                                           sentences  \n",
              "0   Gated - Attention Readers for Text Comprehension  \n",
              "1  A recent trend to measure progress towards mac...  \n",
              "2  Large - scale Simple Question Answering with M...  \n",
              "3  Training large - scale question answering syst...  \n",
              "4  This paper studies the impact of multitask and...  \n",
              "5  However , while most recent efforts have focus...  \n",
              "6  Learning to Compose Task - Specific Tree Struc...  \n",
              "7  In this paper , we propose Gumbel Tree - LSTM ...  \n",
              "8  As an alternative to question answering method...  \n",
              "9  Question answering ( QA ) , which returns exac...  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-393fa051-bdb6-4b91-90c9-c3d1f169897b\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>research_problems</th>\n",
              "      <th>sentences</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Text Comprehension</td>\n",
              "      <td>Gated - Attention Readers for Text Comprehension</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>machine reading</td>\n",
              "      <td>A recent trend to measure progress towards mac...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Large - scale Simple Question Answering</td>\n",
              "      <td>Large - scale Simple Question Answering with M...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>large - scale question answering</td>\n",
              "      <td>Training large - scale question answering syst...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>simple question answering</td>\n",
              "      <td>This paper studies the impact of multitask and...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>Simple Question Answering</td>\n",
              "      <td>However , while most recent efforts have focus...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>Task - Specific Tree Structures</td>\n",
              "      <td>Learning to Compose Task - Specific Tree Struc...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>task - specific tree structures only from plai...</td>\n",
              "      <td>In this paper , we propose Gumbel Tree - LSTM ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>question answering</td>\n",
              "      <td>As an alternative to question answering method...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>Question answering ( QA )</td>\n",
              "      <td>Question answering ( QA ) , which returns exac...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-393fa051-bdb6-4b91-90c9-c3d1f169897b')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-393fa051-bdb6-4b91-90c9-c3d1f169897b button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-393fa051-bdb6-4b91-90c9-c3d1f169897b');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"shape of the training data : \", train_data.shape)\n",
        "print(\"Number of data points in the train data : \" , train_data.shape[0])\n",
        "print(\"Number of feature in the train data : \" , train_data.shape[1])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rUUNL0AekmmX",
        "outputId": "2d77a735-f12d-4bbf-a0b8-13dd88686d42"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "shape of the training data :  (635, 2)\n",
            "Number of data points in the train data :  635\n",
            "Number of feature in the train data :  2\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_data.info()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "l5P5vaexkqQo",
        "outputId": "70a561c4-481c-4a11-be99-b9fccdf114a6"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 635 entries, 0 to 634\n",
            "Data columns (total 2 columns):\n",
            " #   Column             Non-Null Count  Dtype \n",
            "---  ------             --------------  ----- \n",
            " 0   research_problems  635 non-null    object\n",
            " 1   sentences          635 non-null    object\n",
            "dtypes: object(2)\n",
            "memory usage: 10.0+ KB\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_data.head(10)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 322
        },
        "id": "Fy9xUXD1kPrk",
        "outputId": "87ef950f-bcd4-4ff0-b13b-7ef19f8ec8df"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                                research_problems  \\\n",
              "0                                        Grammars   \n",
              "1        Pretraining of Self - attention Networks   \n",
              "2  pretraining a bi-directional transformer model   \n",
              "3                      Language model pretraining   \n",
              "4                            Constituency Parsing   \n",
              "5                  Syntactic constituency parsing   \n",
              "6                            constituency parsing   \n",
              "7                     neural constituency parsing   \n",
              "8          Transition - based Constituent Parsing   \n",
              "9   neural transition - based constituent parsing   \n",
              "\n",
              "                                           sentences  \n",
              "0                  Recurrent Neural Network Grammars  \n",
              "1  Cloze - driven Pretraining of Self - attention...  \n",
              "2  We present a new approach for pretraining a bi...  \n",
              "3  Language model pretraining has recently been s...  \n",
              "4  An Empirical Study of Building a Strong Baseli...  \n",
              "5  Syntactic constituency parsing is a fundamenta...  \n",
              "6  Recent work has proposed several generative ne...  \n",
              "7  Recent work on neural constituency parsing has...  \n",
              "8   In- Order Transition - based Constituent Parsing  \n",
              "9  Both bottom - up and top - down strategies hav...  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-b0a752b6-d776-44e0-ae18-db136ba32913\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>research_problems</th>\n",
              "      <th>sentences</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Grammars</td>\n",
              "      <td>Recurrent Neural Network Grammars</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Pretraining of Self - attention Networks</td>\n",
              "      <td>Cloze - driven Pretraining of Self - attention...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>pretraining a bi-directional transformer model</td>\n",
              "      <td>We present a new approach for pretraining a bi...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Language model pretraining</td>\n",
              "      <td>Language model pretraining has recently been s...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Constituency Parsing</td>\n",
              "      <td>An Empirical Study of Building a Strong Baseli...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>Syntactic constituency parsing</td>\n",
              "      <td>Syntactic constituency parsing is a fundamenta...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>constituency parsing</td>\n",
              "      <td>Recent work has proposed several generative ne...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>neural constituency parsing</td>\n",
              "      <td>Recent work on neural constituency parsing has...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>Transition - based Constituent Parsing</td>\n",
              "      <td>In- Order Transition - based Constituent Parsing</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>neural transition - based constituent parsing</td>\n",
              "      <td>Both bottom - up and top - down strategies hav...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-b0a752b6-d776-44e0-ae18-db136ba32913')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-b0a752b6-d776-44e0-ae18-db136ba32913 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-b0a752b6-d776-44e0-ae18-db136ba32913');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"shape of the test data : \", test_data.shape)\n",
        "print(\"Number of data points in the test data : \" , test_data.shape[0])\n",
        "print(\"Number of feature in the test data : \" , test_data.shape[1])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_NMBokEOkcmi",
        "outputId": "1cc8033e-57c1-4721-be86-8d949e228128"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "shape of the test data :  (330, 2)\n",
            "Number of data points in the test data :  330\n",
            "Number of feature in the test data :  2\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_data.info()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iuNRn73skwTL",
        "outputId": "c8c91edd-17b3-45ba-dfa0-8d1e83f3006e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 330 entries, 0 to 329\n",
            "Data columns (total 2 columns):\n",
            " #   Column             Non-Null Count  Dtype \n",
            "---  ------             --------------  ----- \n",
            " 0   research_problems  330 non-null    object\n",
            " 1   sentences          330 non-null    object\n",
            "dtypes: object(2)\n",
            "memory usage: 5.3+ KB\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# 2. Preprocess the dataset"
      ],
      "metadata": {
        "id": "YBSCj2-2lh5n"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# data cleaning\n",
        "# There were some word which i think might useful in the analysis which are present in the stopword are removed (like dont, would not ,does not , not etc ..)\n",
        "stopwords= set([ 'im','the', 'i', 'me', 'my', 'myself', 'we', 'our', 'ours', 'ourselves', 'you', \"you're\", \"you've\",\\\n",
        "            \"you'll\", \"you'd\", 'your', 'yours', 'yourself', 'yourselves', 'he', 'him', 'his', 'himself', \\\n",
        "            'she', \"she's\", 'her', 'hers', 'herself', 'it', \"it's\", 'its', 'itself', 'they', 'them', 'their',\\\n",
        "            'theirs', 'themselves', 'what', 'which', 'who', 'whom', 'this', 'that', \"that'll\", 'these', 'those', \\\n",
        "            'am', 'is', 'are', 'was', 'were', 'be', 'been', 'being', 'have', 'has', 'had', 'having', 'do', 'does', \\\n",
        "            'did', 'doing', 'a', 'an', 'the', 'and', 'but', 'if', 'or', 'as', 'until', 'while', 'of', \\\n",
        "            'at', 'by', 'for', 'with', 'about','between', 'into', 'through', 'during', 'before', 'after',\\\n",
        "            'above', 'below', 'to', 'from', 'up', 'down', 'in', 'out', 'on', 'off', 'over', 'under', 'further',\\\n",
        "            'then', 'once', 'here', 'there', 'when', 'where', 'why', 'how', 'all', 'any', 'both', 'each', 'few', 'more',\\\n",
        "            'most', 'other', 'some', 'such', 'own', 'same', 'so', 'than', 'too', 'very', \\\n",
        "            's', 't', 'can', 'will', 'just', 'don', \"don't\", 'should', \"should've\", 'now', 'd', 'll', 'm', 'o', 're', \\\n",
        "            've', 'y', 'ain', 'aren', \"aren't\", 'couldn', 'didn', 'doesn', 'hadn',\\\n",
        "            \"hadn't\", 'hasn', \"hasn't\", 'haven', \"haven't\", 'isn', \"isn't\", 'ma', 'mightn', \"mightn't\", 'mustn',\\\n",
        "            \"mustn't\", 'needn', \"needn't\", 'shan', \"shan't\", 'shouldn', 'wasn', \"wasn't\", 'weren', \"weren't\", \\\n",
        "            'won', 'wouldn'])\n",
        "\n",
        "def clean_text(text):\n",
        "    '''Make text lowercase, remove text in square brackets,remove links,remove punctuation\n",
        "    and remove words containing numbers.'''\n",
        "    text = str(text).lower()\n",
        "    text = re.sub('\\[.*?\\]', '', text)\n",
        "    text = re.sub('https?://\\S+|www\\.\\S+', '', text)\n",
        "    text = re.sub('<.*?>+', '', text)\n",
        "    text = re.sub('[%s]' % re.escape(string.punctuation), '', text)\n",
        "    text = re.sub('\\n', '', text)\n",
        "    text = re.sub('\\w*\\d\\w*', '', text)\n",
        "    text = re.sub(r\"(@[a-z]*)\", \"<mention>\", text) #remove any word start with @\n",
        "    text = re.sub(r\"(&[a-z;]*)\", \"<none>\", text) #remove any word start with &\n",
        "    text = re.sub(r\"(#[a-z;]*)\", \"<hash>\", text) #remove any word start with #\n",
        "    text = re.sub(r\"(http|https|ftp|ftps)\\:\\/\\/[a-zA-Z0-9\\-\\.]+\\.[a-zA-Z]{2,3}(\\/\\S*)?\", \"<link>\", text)#remove LINKs\n",
        "    text = re.sub(r'https?://\\S+', '<link>', text) # remove https? links\n",
        "    text = re.sub(r\"(www.[a-z.\\/0-9]*)\", \"<link>\", text)#remove LINKs\n",
        "    text = ' '.join(e.lower() for e in text.split() if e.lower() not in stopwords)\n",
        "    return text"
      ],
      "metadata": {
        "id": "76JO0JAjlgst"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# given each sentence it will call the clean_text()\n",
        "train_data['sentences'] = train_data['sentences'].apply(lambda x:clean_text(x))\n",
        "train_data['research_problems'] = train_data['research_problems'].apply(lambda x:clean_text(x))\n",
        "\n",
        "test_data['sentences'] = test_data['sentences'].apply(lambda x:clean_text(x))\n",
        "test_data['research_problems'] = test_data['research_problems'].apply(lambda x:clean_text(x))"
      ],
      "metadata": {
        "id": "9a97fFTTloDP"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def word_len(text_data):\n",
        "    # create a list to store the length of the sentence\n",
        "    store = []\n",
        "    # iterate thrrough each loop\n",
        "    for i in text_data[:]:\n",
        "        #split each word and count them and append in store\n",
        "        store.append((len(str(i).split())))\n",
        "    return store"
      ],
      "metadata": {
        "id": "pYtgpnDAmiXU"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_data['Number_of_word_in_sentences'] = word_len(train_data['sentences'])\n",
        "train_data[\"Number_of_word_in_research problems\"] = word_len(train_data['research_problems'])\n",
        "train_data[\"difference_word_in_sentences_and_research_problems\"] = train_data['Number_of_word_in_sentences'] - train_data[\"Number_of_word_in_research problems\"]\n",
        "\n",
        "test_data['Number_of_word_in_sentences'] = word_len(test_data['sentences'])\n",
        "test_data[\"Number_of_word_in_research problems\"] = word_len(test_data['research_problems'])\n",
        "test_data[\"difference_word_in_sentences_and_research_problems\"] = test_data['Number_of_word_in_sentences'] - test_data[\"Number_of_word_in_research problems\"]"
      ],
      "metadata": {
        "id": "n7MCrTEUmksT"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_data.head(10)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 620
        },
        "id": "EeXB6b_CnCpu",
        "outputId": "53e7c805-b0d9-4569-e345-9d963a2a12ba"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                                   research_problems  \\\n",
              "0                                 text comprehension   \n",
              "1                                    machine reading   \n",
              "2              large scale simple question answering   \n",
              "3                     large scale question answering   \n",
              "4                          simple question answering   \n",
              "5                          simple question answering   \n",
              "6                      task specific tree structures   \n",
              "7  task specific tree structures only plain text ...   \n",
              "8                                 question answering   \n",
              "9                              question answering qa   \n",
              "\n",
              "                                           sentences  \\\n",
              "0         gated attention readers text comprehension   \n",
              "1  recent trend measure progress towards machine ...   \n",
              "2  large scale simple question answering memory n...   \n",
              "3  training large scale question answering system...   \n",
              "4  paper studies impact multitask transfer learni...   \n",
              "5  however recent efforts focused designing syste...   \n",
              "6     learning compose task specific tree structures   \n",
              "7  paper propose gumbel tree lstm novel tree stru...   \n",
              "8  alternative question answering methods based f...   \n",
              "9  question answering qa returns exact answers ei...   \n",
              "\n",
              "   Number_of_word_in_sentences  Number_of_word_in_research problems  \\\n",
              "0                            5                                    2   \n",
              "1                           14                                    2   \n",
              "2                            7                                    5   \n",
              "3                           17                                    4   \n",
              "4                           26                                    3   \n",
              "5                           32                                    3   \n",
              "6                            6                                    4   \n",
              "7                           24                                    8   \n",
              "8                           26                                    2   \n",
              "9                           26                                    3   \n",
              "\n",
              "   difference_word_in_sentences_and_research_problems  \n",
              "0                                                  3   \n",
              "1                                                 12   \n",
              "2                                                  2   \n",
              "3                                                 13   \n",
              "4                                                 23   \n",
              "5                                                 29   \n",
              "6                                                  2   \n",
              "7                                                 16   \n",
              "8                                                 24   \n",
              "9                                                 23   "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-8adfd1b1-ab61-4176-be91-5ba6d7bbe1f2\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>research_problems</th>\n",
              "      <th>sentences</th>\n",
              "      <th>Number_of_word_in_sentences</th>\n",
              "      <th>Number_of_word_in_research problems</th>\n",
              "      <th>difference_word_in_sentences_and_research_problems</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>text comprehension</td>\n",
              "      <td>gated attention readers text comprehension</td>\n",
              "      <td>5</td>\n",
              "      <td>2</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>machine reading</td>\n",
              "      <td>recent trend measure progress towards machine ...</td>\n",
              "      <td>14</td>\n",
              "      <td>2</td>\n",
              "      <td>12</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>large scale simple question answering</td>\n",
              "      <td>large scale simple question answering memory n...</td>\n",
              "      <td>7</td>\n",
              "      <td>5</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>large scale question answering</td>\n",
              "      <td>training large scale question answering system...</td>\n",
              "      <td>17</td>\n",
              "      <td>4</td>\n",
              "      <td>13</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>simple question answering</td>\n",
              "      <td>paper studies impact multitask transfer learni...</td>\n",
              "      <td>26</td>\n",
              "      <td>3</td>\n",
              "      <td>23</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>simple question answering</td>\n",
              "      <td>however recent efforts focused designing syste...</td>\n",
              "      <td>32</td>\n",
              "      <td>3</td>\n",
              "      <td>29</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>task specific tree structures</td>\n",
              "      <td>learning compose task specific tree structures</td>\n",
              "      <td>6</td>\n",
              "      <td>4</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>task specific tree structures only plain text ...</td>\n",
              "      <td>paper propose gumbel tree lstm novel tree stru...</td>\n",
              "      <td>24</td>\n",
              "      <td>8</td>\n",
              "      <td>16</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>question answering</td>\n",
              "      <td>alternative question answering methods based f...</td>\n",
              "      <td>26</td>\n",
              "      <td>2</td>\n",
              "      <td>24</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>question answering qa</td>\n",
              "      <td>question answering qa returns exact answers ei...</td>\n",
              "      <td>26</td>\n",
              "      <td>3</td>\n",
              "      <td>23</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-8adfd1b1-ab61-4176-be91-5ba6d7bbe1f2')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-8adfd1b1-ab61-4176-be91-5ba6d7bbe1f2 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-8adfd1b1-ab61-4176-be91-5ba6d7bbe1f2');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_data.head(10)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 521
        },
        "id": "aT5OJ8kwHH0u",
        "outputId": "ab133315-0569-4322-ee1b-30978a7c591b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                             research_problems  \\\n",
              "0                                     grammars   \n",
              "1          pretraining self attention networks   \n",
              "2  pretraining bidirectional transformer model   \n",
              "3                   language model pretraining   \n",
              "4                         constituency parsing   \n",
              "5               syntactic constituency parsing   \n",
              "6                         constituency parsing   \n",
              "7                  neural constituency parsing   \n",
              "8         transition based constituent parsing   \n",
              "9  neural transition based constituent parsing   \n",
              "\n",
              "                                           sentences  \\\n",
              "0                  recurrent neural network grammars   \n",
              "1   cloze driven pretraining self attention networks   \n",
              "2  present new approach pretraining bidirectional...   \n",
              "3  language model pretraining recently shown prov...   \n",
              "4  empirical study building strong baseline const...   \n",
              "5  syntactic constituency parsing fundamental pro...   \n",
              "6  recent work proposed several generative neural...   \n",
              "7  recent work neural constituency parsing found ...   \n",
              "8         order transition based constituent parsing   \n",
              "9  bottom top strategies used neural transition b...   \n",
              "\n",
              "   Number_of_word_in_sentences  Number_of_word_in_research problems  \\\n",
              "0                            4                                    1   \n",
              "1                            6                                    4   \n",
              "2                           16                                    4   \n",
              "3                           14                                    3   \n",
              "4                            7                                    2   \n",
              "5                           13                                    3   \n",
              "6                           13                                    2   \n",
              "7                           18                                    3   \n",
              "8                            5                                    4   \n",
              "9                            9                                    5   \n",
              "\n",
              "   difference_word_in_sentences_and_research_problems  \n",
              "0                                                  3   \n",
              "1                                                  2   \n",
              "2                                                 12   \n",
              "3                                                 11   \n",
              "4                                                  5   \n",
              "5                                                 10   \n",
              "6                                                 11   \n",
              "7                                                 15   \n",
              "8                                                  1   \n",
              "9                                                  4   "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-730c278b-62b8-4303-be41-e09fbe9b06c2\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>research_problems</th>\n",
              "      <th>sentences</th>\n",
              "      <th>Number_of_word_in_sentences</th>\n",
              "      <th>Number_of_word_in_research problems</th>\n",
              "      <th>difference_word_in_sentences_and_research_problems</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>grammars</td>\n",
              "      <td>recurrent neural network grammars</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>pretraining self attention networks</td>\n",
              "      <td>cloze driven pretraining self attention networks</td>\n",
              "      <td>6</td>\n",
              "      <td>4</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>pretraining bidirectional transformer model</td>\n",
              "      <td>present new approach pretraining bidirectional...</td>\n",
              "      <td>16</td>\n",
              "      <td>4</td>\n",
              "      <td>12</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>language model pretraining</td>\n",
              "      <td>language model pretraining recently shown prov...</td>\n",
              "      <td>14</td>\n",
              "      <td>3</td>\n",
              "      <td>11</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>constituency parsing</td>\n",
              "      <td>empirical study building strong baseline const...</td>\n",
              "      <td>7</td>\n",
              "      <td>2</td>\n",
              "      <td>5</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>syntactic constituency parsing</td>\n",
              "      <td>syntactic constituency parsing fundamental pro...</td>\n",
              "      <td>13</td>\n",
              "      <td>3</td>\n",
              "      <td>10</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>constituency parsing</td>\n",
              "      <td>recent work proposed several generative neural...</td>\n",
              "      <td>13</td>\n",
              "      <td>2</td>\n",
              "      <td>11</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>neural constituency parsing</td>\n",
              "      <td>recent work neural constituency parsing found ...</td>\n",
              "      <td>18</td>\n",
              "      <td>3</td>\n",
              "      <td>15</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>transition based constituent parsing</td>\n",
              "      <td>order transition based constituent parsing</td>\n",
              "      <td>5</td>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>neural transition based constituent parsing</td>\n",
              "      <td>bottom top strategies used neural transition b...</td>\n",
              "      <td>9</td>\n",
              "      <td>5</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-730c278b-62b8-4303-be41-e09fbe9b06c2')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-730c278b-62b8-4303-be41-e09fbe9b06c2 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-730c278b-62b8-4303-be41-e09fbe9b06c2');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "**Plotting based on word difference**\n"
      ],
      "metadata": {
        "id": "rplZqViBnLiB"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# lets check out the distribution of Number_of_word_in_text and Number_of_word_in_select_text for train data\n",
        "f, ax = plt.subplots(figsize=(12,6))\n",
        "ax1 = sns.kdeplot(train_data['Number_of_word_in_sentences'], shade= True)\n",
        "ax2 = sns.kdeplot(train_data[\"Number_of_word_in_research problems\"],shade= True)\n",
        "plt.title(\"Distribution of Number_of_word_in_sentences and Number_of_word_in_research_problems\")\n",
        "plt.grid()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 405
        },
        "id": "6M6WUzxtnIjC",
        "outputId": "8dc0492e-a9a6-4db6-8d6f-d1301125e42f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 864x432 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# lets check out the distribution of Number_of_word_in_text and Number_of_word_in_select_text for test data\n",
        "f, ax = plt.subplots(figsize=(12,6))\n",
        "ax1 = sns.kdeplot(test_data['Number_of_word_in_sentences'], shade= True)\n",
        "ax2 = sns.kdeplot(test_data[\"Number_of_word_in_research problems\"],shade= True)\n",
        "plt.title(\"Distribution of Number_of_word_in_sentences and Number_of_word_in_research_problems\")\n",
        "plt.grid()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 405
        },
        "id": "Mmwt_dKvHSKE",
        "outputId": "8d0e35d5-f050-424b-d6ba-cd6630a1d799"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 864x432 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Observation from distribution of both the train and test data:\n",
        "\n",
        "\n",
        "*   We got two right skewed plot.\n",
        "*   As we can see the length of sentences are in general more than research problems that is obvious as research problem is subset of the source sentence.\n",
        "*   Similarly there are some over lap too.\n",
        "\n"
      ],
      "metadata": {
        "id": "3M2lysHuoCoz"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# splitting the training data into train and validation set\n",
        "X_train , X_val , y_train , y_val = train_test_split(train_data , train_data.research_problems , test_size = .2,random_state=50)"
      ],
      "metadata": {
        "id": "ZOLR8ELRp7fZ"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def Transfrom_text(textT,textS):\n",
        "    \"\"\" this function will repalce to <tok> which are in both text and selected text\"\"\"\n",
        "    t = \" \".join(\"<tok>\" for i in range (len(textS.split()))) # take each sentence in selected text find the length and put <tok>\n",
        "    textT = textT.replace(textS,t) # replace those word wich are in both text and selected_text with <tok>\n",
        "    return textT # return those sentence\n",
        "\n",
        "def preprocess_text(df):\n",
        "    df['sentences'] = df['sentences'].apply(lambda x: clean_text(x)) # create clean text column\n",
        "    df['research_problems'] = df['research_problems'].apply(lambda x: clean_text(x)) # create clean selected text column\n",
        "    df['target'] = pd.DataFrame([Transfrom_text(df['sentences'][i],df['research_problems'][i]) for i in range(len(df))]) # apply the transform function to all 3 column\n",
        "    return df"
      ],
      "metadata": {
        "id": "Kwx9_4yAqLxM"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_df = X_train.dropna()\n",
        "train_df = train_df.reset_index()\n",
        "\n",
        "val_df = X_val.dropna()\n",
        "val_df = val_df.reset_index()\n",
        "\n",
        "X_test = train_data.dropna()\n",
        "X_test = train_data.reset_index(drop=True)\n",
        "\n",
        "X_train = preprocess_text(train_df)\n",
        "X_val = preprocess_text(val_df)\n",
        "X_test = preprocess_text(X_test)"
      ],
      "metadata": {
        "id": "za6wkwpbqWx3"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "X_train.head(5)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 380
        },
        "id": "e4FwznubsI-R",
        "outputId": "ab260d13-388a-4f31-a3e3-c3de5548dd46"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "   index                           research_problems  \\\n",
              "0    247                                         nli   \n",
              "1    144               recurrent relational networks   \n",
              "2    257  machine reading comprehension mrc based qa   \n",
              "3    151         paraphrase identification questions   \n",
              "4    374                contextual sarcasm detection   \n",
              "\n",
              "                                           sentences  \\\n",
              "0  nli central problem language understanding rec...   \n",
              "1                      recurrent relational networks   \n",
              "2  paper explore task machine reading comprehensi...   \n",
              "3  present solution problem paraphrase identifica...   \n",
              "4  cascade contextual sarcasm detection online di...   \n",
              "\n",
              "   Number_of_word_in_sentences  Number_of_word_in_research problems  \\\n",
              "0                           13                                    1   \n",
              "1                            3                                    3   \n",
              "2                            9                                    6   \n",
              "3                            6                                    3   \n",
              "4                            7                                    3   \n",
              "\n",
              "   difference_word_in_sentences_and_research_problems  \\\n",
              "0                                                 12    \n",
              "1                                                  0    \n",
              "2                                                  3    \n",
              "3                                                  3    \n",
              "4                                                  4    \n",
              "\n",
              "                                              target  \n",
              "0  <tok> central problem language understanding r...  \n",
              "1                                  <tok> <tok> <tok>  \n",
              "2  paper explore task <tok> <tok> <tok> <tok> <to...  \n",
              "3         present solution problem <tok> <tok> <tok>  \n",
              "4  cascade <tok> <tok> <tok> online discussion fo...  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-d7171572-7e14-47cd-ab8d-b2ca036cd3d1\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>research_problems</th>\n",
              "      <th>sentences</th>\n",
              "      <th>Number_of_word_in_sentences</th>\n",
              "      <th>Number_of_word_in_research problems</th>\n",
              "      <th>difference_word_in_sentences_and_research_problems</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>247</td>\n",
              "      <td>nli</td>\n",
              "      <td>nli central problem language understanding rec...</td>\n",
              "      <td>13</td>\n",
              "      <td>1</td>\n",
              "      <td>12</td>\n",
              "      <td>&lt;tok&gt; central problem language understanding r...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>144</td>\n",
              "      <td>recurrent relational networks</td>\n",
              "      <td>recurrent relational networks</td>\n",
              "      <td>3</td>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>&lt;tok&gt; &lt;tok&gt; &lt;tok&gt;</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>257</td>\n",
              "      <td>machine reading comprehension mrc based qa</td>\n",
              "      <td>paper explore task machine reading comprehensi...</td>\n",
              "      <td>9</td>\n",
              "      <td>6</td>\n",
              "      <td>3</td>\n",
              "      <td>paper explore task &lt;tok&gt; &lt;tok&gt; &lt;tok&gt; &lt;tok&gt; &lt;to...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>151</td>\n",
              "      <td>paraphrase identification questions</td>\n",
              "      <td>present solution problem paraphrase identifica...</td>\n",
              "      <td>6</td>\n",
              "      <td>3</td>\n",
              "      <td>3</td>\n",
              "      <td>present solution problem &lt;tok&gt; &lt;tok&gt; &lt;tok&gt;</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>374</td>\n",
              "      <td>contextual sarcasm detection</td>\n",
              "      <td>cascade contextual sarcasm detection online di...</td>\n",
              "      <td>7</td>\n",
              "      <td>3</td>\n",
              "      <td>4</td>\n",
              "      <td>cascade &lt;tok&gt; &lt;tok&gt; &lt;tok&gt; online discussion fo...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-d7171572-7e14-47cd-ab8d-b2ca036cd3d1')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-d7171572-7e14-47cd-ab8d-b2ca036cd3d1 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-d7171572-7e14-47cd-ab8d-b2ca036cd3d1');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "X_test.head(5)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 463
        },
        "id": "znwqz3tkROPl",
        "outputId": "7f0b9fab-d46c-4a27-907c-7d18be217611"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                       research_problems  \\\n",
              "0                     text comprehension   \n",
              "1                        machine reading   \n",
              "2  large scale simple question answering   \n",
              "3         large scale question answering   \n",
              "4              simple question answering   \n",
              "\n",
              "                                           sentences  \\\n",
              "0         gated attention readers text comprehension   \n",
              "1  recent trend measure progress towards machine ...   \n",
              "2  large scale simple question answering memory n...   \n",
              "3  training large scale question answering system...   \n",
              "4  paper studies impact multitask transfer learni...   \n",
              "\n",
              "   Number_of_word_in_sentences  Number_of_word_in_research problems  \\\n",
              "0                            5                                    2   \n",
              "1                           14                                    2   \n",
              "2                            7                                    5   \n",
              "3                           17                                    4   \n",
              "4                           26                                    3   \n",
              "\n",
              "   difference_word_in_sentences_and_research_problems  \\\n",
              "0                                                  3    \n",
              "1                                                 12    \n",
              "2                                                  2    \n",
              "3                                                 13    \n",
              "4                                                 23    \n",
              "\n",
              "                                              target  \n",
              "0                gated attention readers <tok> <tok>  \n",
              "1  recent trend measure progress towards <tok> <t...  \n",
              "2      <tok> <tok> <tok> <tok> <tok> memory networks  \n",
              "3  training <tok> <tok> <tok> <tok> systems compl...  \n",
              "4  paper studies impact multitask transfer learni...  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-6d8e73d5-2603-4180-80e7-d2bd40eaa3b1\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>research_problems</th>\n",
              "      <th>sentences</th>\n",
              "      <th>Number_of_word_in_sentences</th>\n",
              "      <th>Number_of_word_in_research problems</th>\n",
              "      <th>difference_word_in_sentences_and_research_problems</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>text comprehension</td>\n",
              "      <td>gated attention readers text comprehension</td>\n",
              "      <td>5</td>\n",
              "      <td>2</td>\n",
              "      <td>3</td>\n",
              "      <td>gated attention readers &lt;tok&gt; &lt;tok&gt;</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>machine reading</td>\n",
              "      <td>recent trend measure progress towards machine ...</td>\n",
              "      <td>14</td>\n",
              "      <td>2</td>\n",
              "      <td>12</td>\n",
              "      <td>recent trend measure progress towards &lt;tok&gt; &lt;t...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>large scale simple question answering</td>\n",
              "      <td>large scale simple question answering memory n...</td>\n",
              "      <td>7</td>\n",
              "      <td>5</td>\n",
              "      <td>2</td>\n",
              "      <td>&lt;tok&gt; &lt;tok&gt; &lt;tok&gt; &lt;tok&gt; &lt;tok&gt; memory networks</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>large scale question answering</td>\n",
              "      <td>training large scale question answering system...</td>\n",
              "      <td>17</td>\n",
              "      <td>4</td>\n",
              "      <td>13</td>\n",
              "      <td>training &lt;tok&gt; &lt;tok&gt; &lt;tok&gt; &lt;tok&gt; systems compl...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>simple question answering</td>\n",
              "      <td>paper studies impact multitask transfer learni...</td>\n",
              "      <td>26</td>\n",
              "      <td>3</td>\n",
              "      <td>23</td>\n",
              "      <td>paper studies impact multitask transfer learni...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-6d8e73d5-2603-4180-80e7-d2bd40eaa3b1')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-6d8e73d5-2603-4180-80e7-d2bd40eaa3b1 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-6d8e73d5-2603-4180-80e7-d2bd40eaa3b1');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "X_val.head(5)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 413
        },
        "id": "21IzYfYyRSMq",
        "outputId": "b73e364b-79a8-4e26-aded-b21240fd5ef7"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "   index                             research_problems  \\\n",
              "0    203                    directly reading documents   \n",
              "1     82                         question answering qa   \n",
              "2    450  message level topic based sentiment analysis   \n",
              "3    609              neural abstractive summarization   \n",
              "4    571                  temporal relation extraction   \n",
              "\n",
              "                                           sentences  \\\n",
              "0  directly reading documents able answer questio...   \n",
              "1  question answering qa popular topic natural la...   \n",
              "2  datastories task deep lstm attention message l...   \n",
              "3  therefore sequence sequence learning applied n...   \n",
              "4  structured learning approach temporal relation...   \n",
              "\n",
              "   Number_of_word_in_sentences  Number_of_word_in_research problems  \\\n",
              "0                            8                                    3   \n",
              "1                            8                                    3   \n",
              "2                           11                                    6   \n",
              "3                           13                                    3   \n",
              "4                            6                                    3   \n",
              "\n",
              "   difference_word_in_sentences_and_research_problems  \\\n",
              "0                                                  5    \n",
              "1                                                  5    \n",
              "2                                                  5    \n",
              "3                                                 10    \n",
              "4                                                  3    \n",
              "\n",
              "                                              target  \n",
              "0  <tok> <tok> <tok> able answer questions unsolv...  \n",
              "1  <tok> <tok> <tok> popular topic natural langua...  \n",
              "2  datastories task deep lstm attention <tok> <to...  \n",
              "3  therefore sequence sequence learning applied <...  \n",
              "4     structured learning approach <tok> <tok> <tok>  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-c7e8be00-a97d-479e-81b5-7a66d9388846\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>index</th>\n",
              "      <th>research_problems</th>\n",
              "      <th>sentences</th>\n",
              "      <th>Number_of_word_in_sentences</th>\n",
              "      <th>Number_of_word_in_research problems</th>\n",
              "      <th>difference_word_in_sentences_and_research_problems</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>203</td>\n",
              "      <td>directly reading documents</td>\n",
              "      <td>directly reading documents able answer questio...</td>\n",
              "      <td>8</td>\n",
              "      <td>3</td>\n",
              "      <td>5</td>\n",
              "      <td>&lt;tok&gt; &lt;tok&gt; &lt;tok&gt; able answer questions unsolv...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>82</td>\n",
              "      <td>question answering qa</td>\n",
              "      <td>question answering qa popular topic natural la...</td>\n",
              "      <td>8</td>\n",
              "      <td>3</td>\n",
              "      <td>5</td>\n",
              "      <td>&lt;tok&gt; &lt;tok&gt; &lt;tok&gt; popular topic natural langua...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>450</td>\n",
              "      <td>message level topic based sentiment analysis</td>\n",
              "      <td>datastories task deep lstm attention message l...</td>\n",
              "      <td>11</td>\n",
              "      <td>6</td>\n",
              "      <td>5</td>\n",
              "      <td>datastories task deep lstm attention &lt;tok&gt; &lt;to...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>609</td>\n",
              "      <td>neural abstractive summarization</td>\n",
              "      <td>therefore sequence sequence learning applied n...</td>\n",
              "      <td>13</td>\n",
              "      <td>3</td>\n",
              "      <td>10</td>\n",
              "      <td>therefore sequence sequence learning applied &lt;...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>571</td>\n",
              "      <td>temporal relation extraction</td>\n",
              "      <td>structured learning approach temporal relation...</td>\n",
              "      <td>6</td>\n",
              "      <td>3</td>\n",
              "      <td>3</td>\n",
              "      <td>structured learning approach &lt;tok&gt; &lt;tok&gt; &lt;tok&gt;</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-c7e8be00-a97d-479e-81b5-7a66d9388846')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-c7e8be00-a97d-479e-81b5-7a66d9388846 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-c7e8be00-a97d-479e-81b5-7a66d9388846');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "##  below code snippet is for tokenization \n",
        "\n",
        "# now merge all words \n",
        "t2 = pd.DataFrame({'A': X_train.sentences})\n",
        "t3 = pd.DataFrame({'A': X_train.target})\n",
        "# print(t2)\n",
        "all_tokens = pd.concat([t2,t3],axis = 0)\n",
        "\n",
        "# create a tokenizer and fit all the words\n",
        "token = Tokenizer(num_words=54000,filters='')\n",
        "token.fit_on_texts(all_tokens.A) \n",
        "vocab_size = len(token.word_index)+1\n",
        "\n",
        "MAX_LEN = 35\n",
        "# give \"<tok>\" as 1\n",
        "for word,i in token.word_index.items():\n",
        "  if \"<tok>\" in word:\n",
        "    token.word_index[word] = token.word_index[\"<tok>\"]    \n",
        "# it will create a vector based on the sentence word on train text data\n",
        "train_seq_x = token.texts_to_sequences(X_train.sentences)\n",
        "# pre padding\n",
        "train_pad_seq_x = pad_sequences(train_seq_x,maxlen=MAX_LEN)\n",
        "# create vector on train target data\n",
        "\n",
        "# it will create a vector based on the sentence word on train text data\n",
        "val_seq_x = token.texts_to_sequences(X_val.sentences)\n",
        "# pre padding\n",
        "val_pad_seq_x = pad_sequences(val_seq_x,maxlen=MAX_LEN )\n",
        "# create vector on train target data\n",
        "\n",
        "# it will create a vector based on the sentence word on test text data\n",
        "test_seq_x = token.texts_to_sequences(X_test.sentences)\n",
        "# pre padding\n",
        "test_pad_seq_x = pad_sequences(test_seq_x,maxlen=MAX_LEN)\n",
        "\n",
        "# create vector on train target data\n",
        "train_seq_y = token.texts_to_sequences(X_train.target)\n",
        "# # pre padding\n",
        "train_pad_seq_y = pad_sequences(train_seq_y,maxlen=MAX_LEN)\n",
        "\n",
        "# create vector on  target data\n",
        "val_seq_y = token.texts_to_sequences(X_val.target)\n",
        "# # pre padding\n",
        "val_pad_seq_y = pad_sequences(val_seq_y,maxlen=MAX_LEN)\n",
        "\n",
        "# create vector on  target data\n",
        "test_seq_y = token.texts_to_sequences(X_test.target)\n",
        "# # pre padding\n",
        "test_pad_seq_y = pad_sequences(test_seq_y,maxlen=MAX_LEN)\n",
        "\n",
        "# in the target data put 1 and 0 (if tok is present put 1 and all other values to 0)\n",
        "train_pad_seq_y[train_pad_seq_y != token.word_index['<tok>']] = 0\n",
        "train_pad_seq_y[train_pad_seq_y == token.word_index['<tok>']] = 1 \n",
        "\n",
        "val_pad_seq_y[val_pad_seq_y != token.word_index['<tok>']] = 0\n",
        "val_pad_seq_y[val_pad_seq_y == token.word_index['<tok>']] = 1\n",
        "\n",
        "test_pad_seq_y[test_pad_seq_y != token.word_index['<tok>']] = 0\n",
        "test_pad_seq_y[test_pad_seq_y == token.word_index['<tok>']] = 1"
      ],
      "metadata": {
        "id": "shRl0bAzsVoA"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Word embedding with predefined glove"
      ],
      "metadata": {
        "id": "BdO75MoBtV6a"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!wget https://nlp.stanford.edu/data/glove.840B.300d.zip"
      ],
      "metadata": {
        "id": "8zhme3ZOASMo"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!unzip glove.840B.300d.zip && rm glove.840B.300d.zip"
      ],
      "metadata": {
        "id": "nBSeETLKAUl5"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "## using predefined glove vector\n",
        "def embedding_matrix_glove():\n",
        "  # to track the progress of computing\n",
        "  tqdm.pandas()\n",
        "  # create a viable to store glove vector\n",
        "  f = open('/content/drive/MyDrive/ColabNotebooks/glove.840B.300d.txt')\n",
        "  # create a dictionary to \n",
        "  embedding_values = {}\n",
        "  for line in tqdm(f):\n",
        "    # take each word and the vector\n",
        "    value = line.split(' ')\n",
        "    # take the word\n",
        "    word = value[0]\n",
        "    # put all the numeric values to array\n",
        "    coef = np.array(value[1:],dtype = 'float32')\n",
        "    # take the word and array and put together\n",
        "    embedding_values[word] = coef\n",
        "\n",
        "  # take all the arrays from emmbeding values and stacked it\n",
        "  all_embs = np.stack(embedding_values.values())\n",
        "  # perform meand and standard deviation\n",
        "  emb_mean,emb_std = all_embs.mean(), all_embs.std()\n",
        "  #  create an embedding matrix of size (32515 ,300)\n",
        "  embedding_matrix = np.random.normal(emb_mean, emb_std, (vocab_size, 300))\n",
        "\n",
        "  lst = []\n",
        "  # for each word and given count take the embedding values\n",
        "  for word,i in tqdm(token.word_index.items()):\n",
        "      # now take the word get a vector from embedding values(arrays)\n",
        "      values = embedding_values.get(re.sub(r\"[^A-Za-z]\", \"\", word))\n",
        "      if values is not None:\n",
        "        embedding_matrix[i] = values\n",
        "      else:\n",
        "        lst.append(word) \n",
        "\n",
        "  return embedding_matrix "
      ],
      "metadata": {
        "id": "0yNuG0CMsS8s"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "gv  = embedding_matrix_glove()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OlUYCpvgtdey",
        "outputId": "a7371887-3971-4ebc-8c04-5bec99d55c0e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "2196017it [02:15, 16170.37it/s]\n",
            "/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py:2882: FutureWarning: arrays to stack must be passed as a \"sequence\" type such as list or tuple. Support for non-sequence iterables such as generators is deprecated as of NumPy 1.16 and will raise an error in the future.\n",
            "  exec(code_obj, self.user_global_ns, self.user_ns)\n",
            "100%|██████████| 1521/1521 [00:00<00:00, 121063.01it/s]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#3. Machine Learning Model"
      ],
      "metadata": {
        "id": "ANMCogYFypHy"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# refer : https://udibhaskar.github.io/practical-ml/debugging%20nn/neural%20network/overfit/underfit/2020/02/03/Effective_Training_and_Debugging_of_a_Neural_Networks.html#Check-whether-forward-propagation-is-correct-or-not\n",
        "loss_function = tf.keras.losses.BinaryCrossentropy(from_logits=False, reduction=tf.keras.losses.Reduction. SUM_OVER_BATCH_SIZE)\n",
        "\n",
        "##masked loss Eg for sequence output. \n",
        "def maskedLoss(train_pad_seq_y, y_pred):\n",
        "    #getting mask value\n",
        "    mask = tf.math.logical_not(tf.math.equal(train_pad_seq_y, 0))\n",
        "    \n",
        "    #calculating the loss\n",
        "    loss_ = loss_function(train_pad_seq_y, y_pred)\n",
        "    \n",
        "    #converting mask dtype to loss_ dtype\n",
        "    mask = tf.cast(mask, dtype=loss_.dtype)\n",
        "    \n",
        "    #applying the mask to loss\n",
        "    loss_ = loss_*mask\n",
        "    \n",
        "    #getting mean over all the values\n",
        "    loss_ = tf.reduce_sum(loss_)/tf.reduce_sum(mask)\n",
        "    return loss_"
      ],
      "metadata": {
        "id": "i0dj7-DnyfRG"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "##Creating a BLSTM model\n",
        "\n",
        "##fixing numpy RS\n",
        "np.random.seed(42)\n",
        "\n",
        "##fixing tensorflow RS\n",
        "tf.random.set_seed(32)\n",
        "\n",
        "##python RS\n",
        "rn.seed(12)\n",
        "\n",
        "HIDDEN_DIM= 256\n",
        "\n",
        "inputs = Input(shape=(MAX_LEN, ), dtype='float32')\n",
        "\n",
        "embedding_layer = Embedding(vocab_size,300,weights = [gv],trainable = False,name=\"Embedding_layer\")\n",
        "encoder_LSTM_1 = Bidirectional(LSTM(HIDDEN_DIM,return_sequences=True,kernel_regularizer=regularizers.l2(0.01),kernel_initializer=tf.keras.initializers.he_normal(seed=26),recurrent_initializer=tf.keras.initializers.orthogonal(seed=54)))\n",
        "# when the output is a sequence we should use time distributed layer\n",
        "# to understand more please go through this link : https://machinelearningmastery.com/timedistributed-layer-for-long-short-term-memory-networks-in-python/\n",
        "dense_layer_relu = TimeDistributed(Dense(64, activation='relu',kernel_initializer=tf.keras.initializers.he_normal(seed=45)))\n",
        "#dense_layer_relu_1 = TimeDistributed(Dense(64, activation='relu', kernel_initializer=tf.keras.initializers.he_normal(seed=45)))\n",
        "dense_layer = TimeDistributed(Dense(1, activation='sigmoid', kernel_initializer=tf.keras.initializers.glorot_uniform(seed=45)))\n",
        "\n",
        "encoder_embedding = embedding_layer(inputs)\n",
        "Encoded_seq = encoder_LSTM_1(encoder_embedding)\n",
        "\n",
        "outputs = dense_layer_relu(Encoded_seq)\n",
        "outputs = dense_layer(outputs)\n",
        "\n",
        "model = Model(inputs, outputs)\n",
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wpuBv1cQytu8",
        "outputId": "3cd36732-3d93-44ef-9c64-1cce7353d910"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_1 (InputLayer)        [(None, 35)]              0         \n",
            "                                                                 \n",
            " Embedding_layer (Embedding)  (None, 35, 300)          456600    \n",
            "                                                                 \n",
            " bidirectional (Bidirectiona  (None, 35, 512)          1140736   \n",
            " l)                                                              \n",
            "                                                                 \n",
            " time_distributed (TimeDistr  (None, 35, 64)           32832     \n",
            " ibuted)                                                         \n",
            "                                                                 \n",
            " time_distributed_1 (TimeDis  (None, 35, 1)            65        \n",
            " tributed)                                                       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 1,630,233\n",
            "Trainable params: 1,173,633\n",
            "Non-trainable params: 456,600\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "adam = tf.keras.optimizers.Adam(learning_rate=0.0001)\n",
        "model.compile(optimizer=adam,loss = maskedLoss,metrics=['accuracy'])"
      ],
      "metadata": {
        "id": "sufDxJ20yzCD"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plot_model(model)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 466
        },
        "id": "3oNV24QQy3-g",
        "outputId": "070ef487-ace5-4a5e-be43-a412ea54dcee"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<IPython.core.display.Image object>"
            ],
            "image/png": "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\n"
          },
          "metadata": {},
          "execution_count": 36
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import datetime\n",
        "# https://stackoverflow.com/questions/51889378/how-to-use-keras-reducelronplateau\n",
        "def model_check_point_tensor_board():\n",
        "  earlystopper = EarlyStopping(patience=8, verbose=1 ,restore_best_weights = True)\n",
        "  checkpointer = ModelCheckpoint(filepath = 'model_1_1.h5',\n",
        "                               verbose=1,\n",
        "                               save_best_only=True, save_weights_only = True)\n",
        "\n",
        "  reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2,\n",
        "                              patience=2, min_lr=0.00001, verbose=1 , cooldown=1 )\n",
        "  \n",
        "  log_dir=\"logs\\\\fit\\\\\" + datetime.datetime.now().strftime(\"%Y%m%d-%H%M%S\")\n",
        "  tensorboard_callback = TensorBoard(log_dir=log_dir,histogram_freq=1, write_graph=True,write_grads=True)\n",
        "\n",
        "  return [earlystopper , checkpointer , reduce_lr , tensorboard_callback]"
      ],
      "metadata": {
        "id": "Wb7H5NeyN2kM"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# 4. Train the model"
      ],
      "metadata": {
        "id": "LpUBLxHg22Zd"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "history = model.fit(train_pad_seq_x,train_pad_seq_y,batch_size=64,epochs=70,validation_data=(val_pad_seq_x,val_pad_seq_y),callbacks = model_check_point_tensor_board(),verbose=0)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AlrYyWaWy657",
        "outputId": "62a77736-0b90-43bf-8590-3d8a8c07b39a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "WARNING:tensorflow:`write_grads` will be ignored in TensorFlow 2.0 for the `TensorBoard` Callback.\n",
            "\n",
            "Epoch 1: val_loss improved from inf to 0.49048, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 2: val_loss improved from 0.49048 to 0.48588, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 3: val_loss improved from 0.48588 to 0.48418, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 4: val_loss did not improve from 0.48418\n",
            "\n",
            "Epoch 5: val_loss improved from 0.48418 to 0.48246, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 6: val_loss did not improve from 0.48246\n",
            "\n",
            "Epoch 7: val_loss did not improve from 0.48246\n",
            "\n",
            "Epoch 7: ReduceLROnPlateau reducing learning rate to 1e-05.\n",
            "\n",
            "Epoch 8: val_loss did not improve from 0.48246\n",
            "\n",
            "Epoch 9: val_loss improved from 0.48246 to 0.48187, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 10: val_loss improved from 0.48187 to 0.48174, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 11: val_loss did not improve from 0.48174\n",
            "\n",
            "Epoch 12: val_loss improved from 0.48174 to 0.48152, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 13: val_loss improved from 0.48152 to 0.48018, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 14: val_loss did not improve from 0.48018\n",
            "\n",
            "Epoch 15: val_loss did not improve from 0.48018\n",
            "\n",
            "Epoch 16: val_loss improved from 0.48018 to 0.47922, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 17: val_loss did not improve from 0.47922\n",
            "\n",
            "Epoch 18: val_loss improved from 0.47922 to 0.47850, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 19: val_loss did not improve from 0.47850\n",
            "\n",
            "Epoch 20: val_loss improved from 0.47850 to 0.47723, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 21: val_loss did not improve from 0.47723\n",
            "\n",
            "Epoch 22: val_loss did not improve from 0.47723\n",
            "\n",
            "Epoch 23: val_loss did not improve from 0.47723\n",
            "\n",
            "Epoch 24: val_loss improved from 0.47723 to 0.47468, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 25: val_loss did not improve from 0.47468\n",
            "\n",
            "Epoch 26: val_loss did not improve from 0.47468\n",
            "\n",
            "Epoch 27: val_loss did not improve from 0.47468\n",
            "\n",
            "Epoch 28: val_loss improved from 0.47468 to 0.47467, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 29: val_loss did not improve from 0.47467\n",
            "\n",
            "Epoch 30: val_loss did not improve from 0.47467\n",
            "\n",
            "Epoch 31: val_loss improved from 0.47467 to 0.47349, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 32: val_loss improved from 0.47349 to 0.47244, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 33: val_loss did not improve from 0.47244\n",
            "\n",
            "Epoch 34: val_loss did not improve from 0.47244\n",
            "\n",
            "Epoch 35: val_loss improved from 0.47244 to 0.47235, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 36: val_loss did not improve from 0.47235\n",
            "\n",
            "Epoch 37: val_loss improved from 0.47235 to 0.47117, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 38: val_loss improved from 0.47117 to 0.47016, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 39: val_loss did not improve from 0.47016\n",
            "\n",
            "Epoch 40: val_loss did not improve from 0.47016\n",
            "\n",
            "Epoch 41: val_loss did not improve from 0.47016\n",
            "\n",
            "Epoch 42: val_loss improved from 0.47016 to 0.47004, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 43: val_loss improved from 0.47004 to 0.46925, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 44: val_loss did not improve from 0.46925\n",
            "\n",
            "Epoch 45: val_loss did not improve from 0.46925\n",
            "\n",
            "Epoch 46: val_loss improved from 0.46925 to 0.46920, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 47: val_loss improved from 0.46920 to 0.46698, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 48: val_loss did not improve from 0.46698\n",
            "\n",
            "Epoch 49: val_loss did not improve from 0.46698\n",
            "\n",
            "Epoch 50: val_loss improved from 0.46698 to 0.46595, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 51: val_loss did not improve from 0.46595\n",
            "\n",
            "Epoch 52: val_loss improved from 0.46595 to 0.46476, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 53: val_loss did not improve from 0.46476\n",
            "\n",
            "Epoch 54: val_loss improved from 0.46476 to 0.46436, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 55: val_loss did not improve from 0.46436\n",
            "\n",
            "Epoch 56: val_loss did not improve from 0.46436\n",
            "\n",
            "Epoch 57: val_loss improved from 0.46436 to 0.46366, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 58: val_loss did not improve from 0.46366\n",
            "\n",
            "Epoch 59: val_loss improved from 0.46366 to 0.46357, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 60: val_loss improved from 0.46357 to 0.46171, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 61: val_loss did not improve from 0.46171\n",
            "\n",
            "Epoch 62: val_loss did not improve from 0.46171\n",
            "\n",
            "Epoch 63: val_loss improved from 0.46171 to 0.46163, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 64: val_loss did not improve from 0.46163\n",
            "\n",
            "Epoch 65: val_loss improved from 0.46163 to 0.46098, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 66: val_loss did not improve from 0.46098\n",
            "\n",
            "Epoch 67: val_loss improved from 0.46098 to 0.45941, saving model to model_1_1.h5\n",
            "\n",
            "Epoch 68: val_loss did not improve from 0.45941\n",
            "\n",
            "Epoch 69: val_loss did not improve from 0.45941\n",
            "\n",
            "Epoch 70: val_loss improved from 0.45941 to 0.45894, saving model to model_1_1.h5\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['accuracy'], 'r')\n",
        "plt.plot(history.history['val_accuracy'], 'b')\n",
        "plt.legend({'Train accuracy': 'r', 'Validation accuracy':'b'})\n",
        "plt.show()\n",
        "\n",
        "plt.plot(history.history['loss'], 'r')\n",
        "plt.plot(history.history['val_loss'], 'b')\n",
        "plt.legend({'Train Loss': 'r', 'Validation Loss':'b'})\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 513
        },
        "id": "DqJyxdqBOOGB",
        "outputId": "4932865c-3085-4f5a-8aed-c2b8596829d0"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# 4. Evaluation"
      ],
      "metadata": {
        "id": "R2i02iLQNr9r"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# This function decodes the encoded words\n",
        "# takes a variable which will store all the index word\n",
        "reverse_input_char_index = token.index_word\n",
        "def decode_sequence(train_pad_seq_x,input_seq):\n",
        "  # create a storage space to store string\n",
        "  decoded_sentence = \"\"\n",
        "  for i in range(len(input_seq)):\n",
        "    # input seq basically it is the test data\n",
        "    if (input_seq[i] == 1 or input_seq[i] == 2):\n",
        "      if train_pad_seq_x[i] != 0:\n",
        "        # take those sentences which are not cotaining zeros and put in token.index_word which will iteratively decode the words\n",
        "        sampled_char = reverse_input_char_index[train_pad_seq_x[i]]\n",
        "        decoded_sentence += sampled_char + \" \"   \n",
        "  return decoded_sentence"
      ],
      "metadata": {
        "id": "JnZct4LFTdNT"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def jaccard(str1, str2): \n",
        "    a = set(str1.lower().split()) \n",
        "    b = set(str2.lower().split())\n",
        "    if (len(a)==0) & (len(b)==0): return 0.5\n",
        "    c = a.intersection(b)\n",
        "    return float(len(c)) / (len(a) + len(b) - len(c))"
      ],
      "metadata": {
        "id": "mSWoh13RNq45"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def evaluate(pad_seq_x, pad_seq_y):\n",
        "    # after fitting the model this function will be called , and it predict and rounded up\n",
        "    pred = model.predict(pad_seq_x).round().astype(int)\n",
        "    predicted = np.argmax(pred, axis=1)\n",
        "    report = classification_report(np.argmax(pad_seq_y, axis=1), predicted, labels=np.unique(predicted))\n",
        "    print(report)\n",
        "\n",
        "    avg = 0\n",
        "    for i in range(len(pad_seq_x)):  \n",
        "        # now call the decode function , which will give some words and \n",
        "        st1 = decode_sequence(pad_seq_x[i],pred[i])\n",
        "        st2 = decode_sequence(pad_seq_x[i],pad_seq_y[i])\n",
        "        # call the jaccard function to meause the similarity\n",
        "        avg += jaccard(st1,st2)\n",
        "    print(\"Jaccard score = \" , np.sum(avg) / len(pad_seq_x))"
      ],
      "metadata": {
        "id": "dGqX8pYvNlxV"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "evaluate(test_pad_seq_x, test_pad_seq_y)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "r9QJd4sBy-dJ",
        "outputId": "82774279-b667-4421-a5a9-51cfad57a141"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.14      0.60      0.23         5\n",
            "           1       1.00      0.50      0.67         2\n",
            "           2       1.00      1.00      1.00         1\n",
            "           3       1.00      1.00      1.00         1\n",
            "           5       0.40      1.00      0.57         2\n",
            "           6       1.00      1.00      1.00         1\n",
            "           9       1.00      1.00      1.00         1\n",
            "          11       0.75      0.75      0.75         4\n",
            "          12       0.67      0.40      0.50         5\n",
            "          13       1.00      0.80      0.89         5\n",
            "          14       0.56      0.83      0.67         6\n",
            "          15       1.00      0.43      0.60         7\n",
            "          16       0.62      1.00      0.77         5\n",
            "          17       0.90      1.00      0.95         9\n",
            "          18       0.80      0.80      0.80        10\n",
            "          19       0.88      0.82      0.85        17\n",
            "          20       0.79      0.86      0.83        22\n",
            "          21       0.76      0.95      0.84        20\n",
            "          22       0.88      0.88      0.88        17\n",
            "          23       0.70      0.95      0.81        22\n",
            "          24       0.91      0.75      0.82        28\n",
            "          25       0.97      0.76      0.85        42\n",
            "          26       0.71      0.78      0.75        32\n",
            "          27       0.82      0.79      0.80        42\n",
            "          28       0.80      0.76      0.78        42\n",
            "          29       0.84      0.74      0.78        42\n",
            "          30       0.73      0.77      0.75        31\n",
            "          31       0.81      0.75      0.78        61\n",
            "          32       0.85      0.76      0.81        76\n",
            "          33       0.78      0.76      0.77        70\n",
            "          34       0.43      0.60      0.50         5\n",
            "\n",
            "   micro avg       0.78      0.78      0.78       633\n",
            "   macro avg       0.79      0.80      0.77       633\n",
            "weighted avg       0.81      0.78      0.79       633\n",
            "\n",
            "Jaccard score =  0.8366529183852018\n"
          ]
        }
      ]
    }
  ]
}