Tag-based Collaborative Filtering Algorithm Microblog Ecommended

With the vigorous development of Web2.0technology, the global internet has enteredthe network interactive age. The birth and development of microblog network has boughtenormous influence to the transmitting style of the Internet and users’ everyday life. Userscan listen to any topic that they like, be closed to daily life and professional field by usingmicroblog platform. With the rise of the microblog users, the microblog information alsomultiplied growth. Finding the users who have common interests with you in manymicroblog users is just want to get information. Therefore, recommending friends withcommon tastes to microblog users has become the focus of this paper.This paper studied the existing social network friend recommendation algorithm,through summarizing learning recommendation algorithm theory knowledge, and combinedwith the characteristics of microblog users friends, a good friend recommendation algorithmbased on association rules and tag personalized was proposed, which recommended the mostsimilar user to targeted customer for its good friends. Firstly, related concepts of friendsrecommend in microblog were defined, and the friends recommend system implementationprocess was introduced; secondly, calculation methods about common friends relationshipbetween users were presented, which was based on associate rule algorithm, and thencalculation methods about the similarity between the user were presented, which was basedon tag similarity algorithm; finally, co-mbining with common friends relationship and tagsimilarity, the calculation method based on personalized friends recommended was deduced.In the end of the experiment, first the microblog personalized friends recommendsystem was designed, which provided experimental platform support for performancetest of the algorithm. Then weight value test and algorithm performance test were ca-rried out by using good friends recommend system respectively. Thereinto, the weightvalue experimental results show that personalized recommendation algorithm attainedto the optimum when the weight value was0.6. At the last, compared with the existing three friends algorithm, the optimization suitability of personalized friends recom-mendation algorithm was derived, which indicated combined personalized friends recommendation algorithm was effective, and superior to the same kind of friend recom-mendation algorithm in the accuracy and recall ratio.