Yahoo! JAPAN R&D

Have engaged in the research and development in effective information retrieval technologies for Japanese documents, including participations in NTCIRs and TRECs, especially participated in an early epoch of TREC Web tracks and occupied the position of the best performing systems in 2000. Carried out experiments in distributed retrieval, cross language retrieval, named page finding, patent document retrieval and medical document retrieval. Have joined Yahoo Japan Corporation in 2005, and worked on automatic query spell correction, vertical search technologies, automatic classification of e-commerce items and real time clustering tools for service inquiry mails. Meantime, have continued basic research such as document length normalization, pseudo relevance feedback, clustering hypothesis issues, and so on. Participated in the foundation of Yahoo! JAPAN Research in 2007, and works on mining from search user logs and its application to search related services.