The Research And Application Of Searh Engine For Cross-industries Talents

With the rapid development and popularization of the Internet, recruiting from the Internet becomes popular. Research about vertical search engine of professional skills is highly necessary. However, with the increasing number of users in recruitment websites and technology BBS, resume data of users from unknown resource or the BBS users is getting more and more duplicate, which is meaningful to connect discrete domain users.Because it not only simplifies the recruitment works, but also can supplements the user’s personal information.To correlate the resume, first of all, this thesis classifies the resume data set, and then compares the similarity of resumes, the problem of different BBS user’s correlation,is also called multiple networks matching problem. At present, the problem is mainly based on the structure of the network information, using similarity of nodes from different networks to get initial similarity between different network nodes, and then the problem can be transformed into a bipartite graph matching problem,and finally using the KM algorithm to solve the optimal matching problem.In this thesis, according to the characteristics of the existing data, on the basis of the matching algorithm, we improve the calculation method of initial similarity between nodes. Mainly because there are more than two layers of the networks’ structure relations, and the network nodes(users) still have a lot of attribute information, such as posting record and labels, etc., taking advantage of the attribute information can optimize the similarity calculation between users. This thesis also uses the cache skill to improve the indexing of the search engine, using synonym words of skills to improve the search method, optimize of fuzzy search, and connect with the actual situation to sort in order to search the results more in line with people’s needs.The analysis of search results indicates that the improved search method can enrich results,and the improved sorting preferences will be more in accordance with the users’ expectations. Through comparing our algorithm with structured data from Zhihu and Weibo, we find that the improvement of similarity of nodes will meliorate the search results.