text-mining

Classes to explore various text embeddings and learning models

Vectorizer:

• custom / from-scratch TF-IDF matrix encoding
• custom weights to improve matching

Text Classifier:

• logistic regression on TFIDF with custom tokenization
• comparison of training/testing on two different datasets -- seeing how well it generalizes
• testing model on custom vocabulary created from intersection of the two datasets

Continuous Embeddings

• comparison of logistic regression on basic TFIDF embeddings to doc2vec embeddings
• seeing how well model generalizes over different test-train splits