Korean Information Retrieval Model Based on the Principles of Word Formation

This paper presents an information retrieval model based on the principles of word formation to alleviate the mismatch between index terms and query terms. The model provides the partial matching method using morphological information which represent how morphemes consist of words. In this paper, we suggest a eecient method of dealing with derived words and compound words. The contribution of this research is the successful recovery against the diierence of surface forms between index terms and query terms. The eeectiveness of using partial matching for conventional indexing methods is evaluated with KTSET 2.0. The experimental result shows that partial matching consistently improves retrieval performance.