Research On The Key Techniques Of Contextual Search

Traditional Internet search has made very good effect on the recall level of information research, but it is difficult to meet the user’s individualized demands and can’t get the search results varied from person to person and can’t provide personalized result based on one’s contextual information.This paper is focused on the key technology of contextual search. We find that most of the contextual search either at home or at abroad only take use of position and time information to provide personalized service without mixing together each other and other context during one’s research, so the search results can’t meet one’s individualized demand very well. If we add one’s social situation to search engine, and combine the various personalized technology, it is bound to bring more personalized service.Regard of this point, we first make a brief introduction of search engine’s structure and personalized service, then we conduct research on social network analysis, expansion query, the acquisition of context, personalized recommendation and other related technology. After that, we propose a hybrid model based on the contextual search on mobile platform. With a perspective on three angles (subject, object, subject relations) for search modeling, we build user profile based on the yellow pages’feature and action history. We also build a social graph for each person based on his/her social relations. We will increase novelty and credibility of the search results on the basis of social graph, and add results to search based on the personalized recommendation algorithm. After all these work have conducted, we build a hybrid model by using a linear-weighted formula. On the basis of social map, we can increase the novelty and credibility of search results. In this way, we use personalized recommendation algorithm can bring good performance. After all the above models have done, we lead to a context-based search fusion model based on linear weighted fusion.Then, we design a contextual search system based on mobile platform and the hybrid model algorithm will be applied to the system design. We propose a set of key indicators of the evaluation of the contextual search system. In the subsequent experimental data, more than3000users and460,000yellow pages are involved, and a lot of testing through three aspects will be conducted. The results of test indicate that the hybrid model algorithm in contextual search system is feasible. Finally, we summarize the current work, and propose a couple of deficiencies and the future work we should do out of this paper.