The Study Of Reciprocal Recommendation Algorithm For The Field Of Recruitmenty

The traditional personalized recommender system recommends information andproducts which are interested by users to them based on their interest characteristic andbuying behavior. But, as the expanding of recommendation requirement, in the particularfields which involves the matching between people and people, in theserecommendations, both sides are the subjects what have the right of free choice, and theyhave to satisfy both requirements when matching. Therefore, in this recommender systemhow to solve the reciprocal recommendation becomes an urgent question. In this paper,based on the comprehensive analysis of research status at home and abroad, we have afurther study on reciprocal recommendation algorithm.Firstly, according to the problem that traditional recommendation algorithm can notmeet the preference of both users, we introduce the concept of ontology and propose aontology-based reciprocal recommendation model, based on the depth that when domainontology describe information, the ontology-based user model and the similaritycomputing method are proposed, the bidirectional matching-based reciprocalrecommendation is proposed according to the bipartite graph in mathematical modeling,at last, it realizes the two-way choice in recommender system.Secondly, in view of the network recruitment system, we analysis the feature of userpreference model， a implicit preference calculation method based on users’communication history is proposed, it excavates their potential preference information,overcomes the defects of existing preference function and improves the accuracy ofdescribing user information. And from the aspect of user’s similarity computation, theuser’s explicit and implicit similarity is integrated by introducing the weight value α, itcan reflect the true association between users much more comprehensive; in addition, inorder to achieve the reciprocal in the recommender system, the recommendation processis divided into two single parts, we have to recommend them to each other Respectively,satisfy their needs and get the result of reciprocal.Finally, we have to verify the algorithm and contract it to the existing reciprocal recommendation algorithms, summarize and look forward to the future researchdirection.