jonathandunn/text_analytics

Provides unified APIs for feature extraction (style n-grams, sentiment, TF-IDF content, syntactic constructions), supervised classification via SVM and MLPs, and unsupervised methods including LDA topic modeling, Word2Vec embeddings, and document clustering. Built as a pedagogical toolkit accompanying free edX courses, it combines traditional linguistic features with neural approaches in a single TextAnalytics class that handles preprocessing, phrase detection via PMI and Delta P association metrics, and corpus comparison operations.

121 stars. No commits in the last 6 months.