Natural Language Text Retrieval Using a Large Semantic Network

摘要

The primary objective of this SBIR effort is to explore and demonstrate advancedperformance in precision and recall for text retrieval software. Advanced performance will be accomplished by using dictionaries (specifically WordNet from Princeton University) and other enhancements to the base technology of ConQuest (improved ranking, syntactic processing, etc). Demonstration of accuracy will be performed as a series of retrieval experiments and tests, using the TREC-2 corpus and queries.