Research On Location-Based Mobile Web Search Method

With the popularization of mobile device and evolution of wireless network, mobilesearch has been expanding its quotient in search market, and began to be an importantdomain of search research area. Current mobile search applications mostly copy websearch engines, or just simply use location and distance to rank result set. This rankingalgorithm can’t reflect the actual distance between users and targets precisely, causinginconvenience in the users’ access to near targets. As a result, users’ search experiencedeteriorates.This paper describes a new page ranking algorithm of mobile web search engine todetail the mobile search conditions as introducing location to mobile web search. There istwo aspects in calculating and using location information in mobile web search research,how to extract location information and how to use location information. This paper isbased on these two aspects. First, extract the geographic entity names from web pagecontent using CRF(Conditional Random Fields), disambiguate them with established rulesfor Chinese geo-names, and eliminate geo-names’ distractions using Decision Treealgorithm. Then, this paper uses a comprehensive ranking algorithm combined with pages’importance and location information to re-rank the preliminary result set provided byordinary web search engines, in order to get a higher relevant score of top N results to liftmobile users’ search experience.This paper also designs some experiments to evaluate two aspects’ research.Experiments’ results show that, this page content location extraction combined with CRF,established rules and Decision Tree, has a good performance on average precision andrecall, this comprehensive re-ranking algorithm gets an improve in precision.