Skip to content

Project for Natural Language Processing Course at Lakehead University

Notifications You must be signed in to change notification settings

vatsal2727/Google-Smart-Compose

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 

Repository files navigation

Smart Compose - Natural Langugage Processing

This Project discusses topic of sentence or word autocomplete, which is in world of Natural Language Processing also known as Language Modelling. Idea of this project is inspired by Smart Compose provided Google G-mail and mobile phone keyboards. Language Modelling uses NLP and machine learning to interactively offer sentence completion suggestions as user type anything, allowing user to draft sentence or text faster and efficiently. In our model suggestion are usually made based on the training corpus text. This application can potentially reduce grammar errors as well help in sentence formation. Paper includes Importance of this application and NLP, Dataset, Architecture, Performance, Literature Review, pros and cons of using this application and GUI component.

There are many application of NLP and since most of them are somehow related to human language and text, and voice. It is one of the most widespread used application is sentence auto-complete or suggestion. This application uses concept of Language Modeling. A trained language model learns the likelihood of occurrence of a word based on the previous sequence of words used in the text. So its a sequence to sequence modeling. Auto-complete is used at many places like Google search engine, G-mail, Mobile Key-pad and many more. Smart compose by G-mail makes prediction based on current email context, mail they are replaying and subject body. Google also uses the large dataset of emails to train the model. And users will be suggested subsequent word while they type. We have used NLP concepts such as n-gram, char-gram, data pre-processing, Language Modelling and probabilistic distribution, Tokenization, word embedding, Recurrent Neural Network, RNN with Long Short Term Memory (LSTM).

About

Project for Natural Language Processing Course at Lakehead University

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published