Research On Topic Optimization Integrated Collaborative Recommendation For Social Tagging

Social tagging system is a mobile social application platform,where users are becoming more and more active.With the continuous increase of users,resources and tags,the system gradually presents the characteristics of “big data” such as large number,fast growth,complexity,and unreliable quality,which greatly increases the complexity of recommendation.The contradiction between efficiency and effectiveness of recommendation service in social tagging is increasingly becoming prominent.Therefore,combined with the National Natural Science Foundation of China(NSFC)"Research on social-driven context-aware personalized information service in ubiquitous computing environment"(Project No.71471165),this paper presents incorprating topic optimization into collaborative recommendation for social tagging,which alleviates the problem of redundancy,uncertainty and inconsistency existing in “big data”.The main contributions of this thesis are listed as follows:(1)A recommendation service model integrated optimization process is proposed.Combining the idea of optimization before service,the process of recommendation implementation in social tagging is divided into two processes: off-line topic optimization and on-line recommendation service.By integrating of the two processes,personalized recommendation service with high quality and efficiency can be achieved.(2)The improved LDA model is used to optimize the tag topics.Based on the “user-tag” binary relationship,the improved LDA model is established by utilizing the characteristics of annotation behavior to mine potential tag topics.By dividing the original cluttered tags into clusters with consistent topics,the problem of the redundancy,uncertainty and inconsistency in tags can be eliminated.(3)A collaborative recommendation method for integrated topic optimization is proposed.The “user-topic” preference model and the “resource-topic” weight model are established by off-line topic optimization.Finally,a user preference model for resources under multi-interest topics is constructed to improve the efficiency of on-line recommendation.(4)Experimental exploration is carried out from Movielens.The quality of recommendation is measured by accuracy,recall,F-measure and diversity,and the efficiency of recommendation is measured by time complexity and recommendation generation time.While obtaining the optimal parameters of the recommendation method,a number of comparative experiments are set up to explore the balance effect of the recommendation method on recommendation quality and efficiency.The research shows that the proposed collaborative recommendation method for integrated topic optimization is superior to the traditional collaborative recommendation method,especially in the diversity index of recommendation.In addition,our method effectively improves the on-line recommendation efficiency visible to users,and achieves the balance of recommendation quality and efficiency.The results of this research have great practical value for high-quality and efficient recommendation service.