Statistical Semantic Discovery

Automatically discovers semantic relationships by analyzing value co-occurrence patterns across table corpora. Handles entity variations, hierarchical relationships, and code mappings without manual rules.

PMI-Based Join Quality

Uses Pointwise Mutual Information scores to quantify relationship strength. Two proven algorithms: CS-JP-LP for optimal quality and RS-JP for efficient performance with quality superior to traditional approaches.

Corpus-Driven Intelligence

Based on Microsoft Research paper. Stores co-occurrence patterns from your table corpus, then calculates semantic relationships on-demand for fast joins that understand your data domain without external knowledge bases.