TIPSTER Phase I Final Report

Overview: During Phase I of the TIPSTER program, HNC developed a unique approach to machine learning of similarity of meaning. This approach, embodied in a system called "MatchPlus", exploits this learned similarity of meaning for concept-based text retrieval, routing and visualization of textual information. MatchPlus uses an information representation scheme called "context vectors" to encode similarity of usage. Key attributes of the context vector approach are as follows: