Word2Vec Experiment
Some quick "Getting Started" tools for a Word2Vec experiment.
train.py
Train a Word2Vec model on some input corpus. Expects entire corpus to be concatenated into one file with one sentence per line.
Takes two arguments - input corpus file and output model file
interact.py
Load a model and allow the user to query for terms that are similar to input. Type "\quit" to exit the program.
Takes one argument: The model from which similarities are derived
evaluate.py
Loads two models and computes the amount of agreement between their representation of terms using Average Jaccard Similarity
Takes two arguments: Two Word2Vec models to be used for evaluation