Research On Weibo Recommendation Algorithm Based On Domain Characteristics And User Preferences

With the popularity and rapid development of social networks,more and more users pass through the social network to the outside world,and get information in real time.As a typical social network,micro-blog attracts a large number of users for its short,real-time and easy-tooperate features.The huge amount of users produced explosive information on social network.Currently,the research hotspot on social network research is to mine users' preferences from data information generated by users and recommend possible micro-blogging to users.The paper is based on the existing problems to realize the recommendation of the micro-blog system.First of all,there is a problem that the existing recommendation algorithm does not consider the individual analysis of the user's social network,this paper proposed an algorithm based on fusion of the domain features and improved K-Shell micro-blogging user influence analysis;then,in order to solve the problem of the recommendation is not high enough and result is not personalized,this paper proposed a personalized micro-blogging recommendation that fuses user features and preferences.Finally,using the above two algorithms to design and implement a micro-blogging recommended prototype system based on domain features and user preferences.The specific research work as follows:(1)Proposed the characteristics of the fusion field and improved the K-Shell micro-blog user influence analysis method.Firstly,this artical proposed two methods of user domain partitioning: the algorithm based on domain classification microblog and based on domain division based on social relationship.according to the user's characteristics to seleceted the correspond domain division;Then,improved the Ks value of the node in the K-Shell algorithm by the importance degree of the similarity of the second nearest neighbor.Finally,the value of the domain influence of the user is derived in combination with the user field and the importance of the user.Through the experimental verification,the accuracy of calculating the user influence value is higher than the traditional K-Shell algorithm and other algorithms,and the effect is better.(2)Proposed personalized micro-blogging recommendation that fuses user features and preferences.Firstly,this article introduced the time effect function and used the three characteristics of the user to calculated user similarity,and filtered out the user's nearest neighbor.Secondly,calculate the user's real-time preference for different categories to obtain the user's preference category.In the end,the user's nearest neighbor's favorite micro-blogging is composed of the same micro-blogging as the user's preference category,which makes up the recommendation list,and calculated the authoritative value of the micro-blogging in the recommended list,and predicted the degree of interest of users on micro-blogging based on the improved prediction scoring formula,and selected the micro-blogging top-k to recommend users according to the predicted value.(3)Designed and implemented Micro-blogging recommended prototype system based on domain features and user preference.In this paper,based on the micro-blogging user influence analysis and the personalized micro-blogging approach,this article analyzed and designed the modules and processes that are required in the system,and ultimately implemented the micro-blogging which based on the field characteristics and preferences,it can be used to make the tweets that the user might be interested in based on the user characteristics.