MEDICAL DOMAIN OPTIMIZED QUERY EXPANSION

摘要

The present invention relates to a query expansion technology specified in a medical domain, which automatically adds additionally related query words to a directly inputted query word when a user inputs the query word in a retrieval system, thereby enhancing retrieval quality. A query expansion method considering a medical domain knowledge structure and a retrieval intention comprises the steps of: semi-automatically constructing a query expansion word dictionary for each medical domain knowledge type from corpora; classifying co-occurrence information between a medical domain query word and words by the medical domain knowledge type and performing calculation; expanding the medical domain query word into words of the type of ″disease name″, ″symptom″, ″cause″ and ″treatment method″ in balance to show good performance in a general retrieval scenario of the medical domain; and filtering the expanded words considering the relationship (co-occurrence information) among the expanded words.