SSAT Analogy Solver

I gave a lightning talk presentation on this project. Here is the link to the slides.

Background info

• Load vector embeddings from GloVe
• Solver
• Testing on real questions
• Web scrapping for gathering mass data on questions and answers

Data Collecting:

• Web scraping

Data Processing:

• Putting the questions scrapped from data to ssatQuestion class.
• Putting the records into CSV File

Data Cleaning:

• Remove all incomplete questions: missing one of the three: Question Prompt, Answer Choice, Correct Choice

Data Analysis:

• Feeding the questions into solver using GloVe vector embedding and Gensim library functions, test the accuracy in regard to 1. Training data size, 2. Vector dimension embedded.

Here for training your own embedding using word2vec and explaining underlying reasoning for analogies in vector semantic.