2009-Named entity recognition in query教程.pdfVIP

Named Entity Recognition in Query † ‡ † ‡ Jiafeng Guo , Gu Xu , Xueqi Cheng , Hang Li † Institute of Computing Technology, CAS ‡ Microsoft Research Asia Beijing, P.R.China Beijing, P.R.China guojiafeng@,cxq@ {guxu, hangli}@ ABSTRACT entity and assign “Game” to it as the most likely class, This paper addresses the problem of Named Entity Recog- “Movie” and “Book” as less likely classes, and “Music” as nition in Query (NERQ), which involves detection of the unlikely class. This is because the context “walkthrough” named entity in a given query and classification of the named strongly indicates that “harry potter” here is more likely to entity into predefined classes. NERQ is potentially useful in mean the Harry Potter game. (If the query is only “harry many applications in web search. The paper proposes tak- potter”, then “Book” and “Movie” will be more plausible.) NERQ is essentially useful for many applications in web ing a probabilistic approach to the task using query log data and Latent Dirichlet Allocation. We consider contexts of a search. According to our analysis, about 71% of search named entity (i.e., the remainders of queries after the named queries contain named entities. Identifying named entities entity is removed) as words