Research On Intellisensing Of User Preferences Based On Bookmark Social Network

To begin with, personalized recommendation and public sentiment focus oninformation retrieval, information similarity, and information accumulate, and key nodeof information as well. The ability of deal with these four questions can determine theability of personalized recommendation system and public sentiment system, which hasbeen widely researched nowadays. The paper abstracts the URL and Users as nodewhich has diversity interests, and then the research work can be recognized that miningcharacteristics and concepts from large-scale dynamic complex social network.Furthermore, the main difference between traditional recommendation methodsand this paper is completely adopting user behaviors, and adopting tags to miningURL’s concepts and Users’ preference. Finally, the paper uses single user behaviors andgroup user relationship to analysis and gain regular pattern from bookmark socialnetwork.In addition, the work analysis sign in and log out of URL which in bookmarksocial network, at the same time, the research employs math analysis and data mining asresearch tools, the study is concerned with social bookmark system under time andspace environment, and the work reveals that the construction and analysis of socialbookmark system which adopts innovation in the scientific approaches. The aim of theresearch is to accomplish intelligent sense for preference of user interest based on userbehavior data. The main conclusion of this research includes:(1) Adopting Interest Feature Spatial Model to solve the intelligent sense of userinterest and interest trend in social bookmark system. We propose a new spatial indexstructure of IFS-Tree. According to the concept similarity based on HowNet,computing formula of user interest, objects location in the IFS, the construction anddynamic regulation of IFS-Tree, and computing formula of user interest similarity. Aside from solving the recommendation problems of current user interest, we alsopropose short-term interest, real-time interest, potential interest and user interesttrending, and propose six recommendation strategies.(2) In order to solve the problem of user relationship and group division in socialbookmark system, this paper has been proposed MSFP based on SFP. In order to savethe insert time, giving ascending sort head list and core-node link. At the same time,pruning strategies have been used in AFP-Tree, CMP-Tree, SFP-Tree and CFP-Tree.(3) We propose a novel method of mining subgroup in social network based onMSFP (Maximal Sequent Frequent Pattern). Firstly, giving the computing formula ofobject relationship in IFS, and then proposing the generating of adjacency matrix insocial network and dynamic regulation strategy. Secondly, proposing the conception ofsets of weights K accessibility nodes, and also proposing the conception of MSFP andmining algorithm of MSFP and the algorithm of discovering subgroup.(4) The application of this research is mainly on effective personalizedrecommendation area, and the research focus on user behavior information retrieval andintelligent sensor-network of user interest. For metrics of real-time, precise,comprehensiveness, fashionable, diversity and freshness in personalizedrecommendation, we construct novel personalized recommendation based on userinterest feature, user relationship and user behavior. The performance demonstrates thatIFS can fulfill millisecond response, and MSFP can efficiently find user group whichhas the same interest, some works of social network tree are still testing.