This post introduces a feature selection algorithm and Neo4j extension that can select on the most relevant features extracted from a corpus of text and be used for high-performance sentiment analysis on movie reviews.