Research On The Knowledge On Organization Pattern Based On Users In Social Network

Currently,most information on the social networking are superficially organized according to classification navigation and information retrieval.Obviously,efficiency of users' information searching behaviors are affected due to the huge amount of data.However,if we can make user-centered organizations of the user generated content and knowledge community,information retrieval efficiency can be improved.Thus,it's significant to study about user knowledge organization.In this paper,the hierarchical topic system construction of user generated content,user modeling,community detection and the association among these three aspects are studied.We use Weibo as platform,and take online social networking users as research objects.The hierarchical topic system construction of user generated content can make users know the relationship of each topic quickly and find their interested contents accurately.The topic model is firstly used to mine the potential topics and extract document collection under each topic.And it is applied again to mine sub-topic based on the first results.Finally,the hierarchical topic system is constructed to understand the topic content and topic heat.In the part of user modeling,microblogs which are users' original ones,retweeted ones and liked ones are combined differently,and the author topic model is used on each data set to build user's own model separately.User interest model is also updated by adding models of follower collection and followee collection.In addition,institutional users are studied and the hot topics and topic evolutions of them are revealed by adding time dimension.In the part of community detection,label propagation algorithm is used to detect community on follower relation data,followee relation data and the mixed data of them.A good community partition results are obtained then.Combining with user information,composition of each community is analyzed.It is obviously that different communities have different topics.When building of user-centered multi-dimensional knowledge system,the hierarchical topic system,users and communities are combined by computing similarities between each other.Users can get information from multiple perspectives and the information acquisition efficiency is improved.The meanings of this work are as follows.The results of user modeling above all can solve data sparse problems and discover users' interests more comprehensively.Moreover,knowledge construction of user generated content,community detection and user-centered multi-dimensional knowledge system can be applied in information recommendation field to recommend content,friends and communities to users which they are interested in.The study of user-centered multi-dimensional knowledge system can provide references for current social networking websites,such as information organization and user interface design.