Information Retrieval Based on Linguistic Structure

Takashi Miyata

This talk introduces an information retrieval system which utilizes linguistic structure of documents and queries. Although previous researches have reported that costly structural matching does not contribute to the improvement of `efficiency' in information retrieval, we noticed that structural information could improvisationally provide useful hints for users to revise their queries. We would like to report our prototype system, which is currently evaluated with some subjects on retrieval tasks for newspaper articles.