Recommendation Research Based On User Activity And Hot Topics In Micro-blogs Community

Microblog including Facebook, Twitter and Sina Microblog attracts massive users andprovides people a broad platform, on which people can share information within their virtualcommunities anywhere at any time. User relationships in microblog are not mostlysymmetrical, because a user can have a lot of fans, but the user does not need to reverse thepast focus on these fans, users have the initiative to choose and decide their behavior ratherthan passively action. As the impact of asymmetric relations, microblog is a broadcastcommunication medium where information dissemination is in large scale involvingmulti-node interactions. However, it is very crucial to how to access to the most valuableinformation fast and efficiently is very important.A recommendation in micro-blogs community can be thought of as a filtering andranking system for contents updated during a given period, which gets and recommends thelatest and most useful information to users. Recently many domestic and foreign scholars putforward the microblogging community models and the microblogging communityrecommendation algorithms to improve the precision of community recommendation andachieve some results. But with the continuous development of network technology and thecontinuous renewal of social tools, the way and platform that people express their views isdiverse, social practices also changes gradually, which make the previous traditionalmicroblogging community recommended methods are insufficient to meet the needs of newsocial network. Based on the above background, this thesis first starts from the basic theory ofthe microblogging community, analyze the existing community recommendation technology,and advances the microblogging community recommendation method based on user activityand hot topics. The main research contents are list of as follows:(1) On the basis of the existing Web community model, this thesis analyzes the behaviorrelationship in the microbog community and the range of community communication, buildsthe behavior model and the communication process model, and at the same time analyses thecharacteristics of the microblog community.(2) Based on the analysis of the activity of microblog users, the thesis gives therelationships between the number of Tweets and the number of users, and the relationshipsbetween the number of tweets and the number of fans relationships.(3) Based on each user’s interests and the concerns relationship between users, the thesiscombines a key role of the time factor in the detection of hot topics to study how toeffectively get the latest and most valuable topic issues in the community, and proposes ahot topic detection algorithm with the consideration of the decay of time parameter based onPageRank algorithm. (4) Microblogging community recommendation algorithm which recommends thosepeople which are active and publish hot topics to new users in the community and improvesthe retrieval speed and efficiency of new users that can quickly and accurately access toinformation is proposed based on user activity and hot topics.(5) The experimental results over real data set illustrate the effectiveness and efficiencyprovided by our algorithm.