CSCI381-780 Assignment 3-Word2vec Solved

1 Word2vec

In this assignment, you will use a training corpus news.crawl to train Word2Vec. You should follow the instructions found at: https://kavita-ganesan.com/gensim-word2vec- tutorial-starter-code/#.XqO4C2ZKj3h
Please use the following parameters:

1. size=150
2. window=5
3. min count=2 4. iter=10

Questions:

1

1. Report similarity scores for the following pairs: (dirty, clean), (big, dirty),(big, large) , (big,small)
2. Report 5 most similar items and the scores to ’polite’, ’orange’
3. Now change the parameters of your model, as follows: window=2, size=50. Answer the 2 questions above for this new model.