Design And Implementation Of Weibo Following Recommendation System Based On User Relations Chain

The Weibo service is a new application of Web2.0technologies, to provideconvenient, fast content distribution and sharing service for the Internet users, becomesone of the important applications for the Internet users in the daily life. However, with therapid development in recent years, the Weibo platform goes into the era of informationoverload, and therefore a effective Weibo recommendation system has a key role inobtaining the relevant information timely for general readers. Although, the mainstreamrecommendation methods (such as Facebook, Twitter and Sina Weibo usersrecommendation model) use the content-based recommendation method, theneighbors-based recommendation method, the user interest-based recommendationmethod and so on, these methods exist some problems, like the relationships’ density ofsocial network is not enough, the recommendation relationship is limited, social awarenessis superficial and the process of building relationships is slow etc..Established a new Weibo following relational recommendation model-relationalchain model, including Weibo social network model of strong ties and weak ties. Formerdefinition once the relationship type, the weights of each type of relationship based onmachine learning and experience, a comprehensive matrix calculated once the relationship,and then once the relationship on the basis of second degree relationship matrix derivedrelationship matrix corresponding parameters do multiplication calculated summing the16kinds of the second time relationship, and obtaining the current user’s recommendationthat users result sets, after sorting and filtering conditions, the final result isrecommendation to the user. The latter is the first user to follow relations (followers)similarity based on the similarity of the the idol (follow objects) between users and fans toget interested again in accordance with the demographic attributes and mining interest characteristic dimension similarity, integrated by users similar matrix given userrecommendation results based on the similarity matrix.Tencent Weibo platform forcross-platform data and its own platform data within the acceptable range of user privacy,the relationship chain recommendation algorithm, recommendation following list inaccordance with the the scenes demand to users.And achieve the recommendation results evaluation, that recommendation systemcan provide to customer satisfaction and reasonable recommendation results come throughthe online testing method AB testing different barrels following data, comparing therecommendation results can improve the operation of the user’s following, showingRecommendation system advantages. Meanwhile, the per capita chain of the wholeamount of user data, derived data showing rising in the last line, the situation, andrecommend results reasonable to recommend to the user to follow to the object, reducingthe user to take the initiative to build the relationship between the threshold of the chain,to facilitate the formation of a user’s core relationship chain, providing user activity, toenhance Weibo application of the word-of-mouth.