Research On The Influence Of Network Social Users And Friend Recommendation

With the rapid development of social networks,online social networking platforms have become the mainstream information exchange channels for modern people.First of all,as the metric of user information dissemination ability,network user influence is widely used in social hot news mining and public opinion orientation research fields.Researching social network user influence helps to explore the key path and key nodes of network information dissemination.How relevant departments can accelerate the dissemination of positive information and control the diffusion of negative information to provide decision support.Secondly,through the current mainstream social network platform,users can not only get relevant help and information of interest,but also get to know more friends and expand their circle of friends.These are very meaningful.The main work of this paper has the following two points.Firstly,the traditional PageRank algorithm adopts the mean distribution method to cause the PR value loss problem.A parameter-adjustable social user influence measurement method is proposed.The method first considers the PageRank algorithm PR value distribution law and analyzes the user behavior frequency.The corresponding adjustable parameters are calculated,and then the adjusted parameters are combined with the calculation of the influence formula of the user interaction behavior to obtain the adjusted PR value.Finally,the influence of the user's social relationship is considered,and the social network user influence measurement formula is comprehensively obtained.The algorithm is compared with other algorithms on the same data set,and the calculated RMSE value is the smallest,which indicates that the user influence ranking result obtained by the algorithm is closer to the real user influence ranking.Second,as mobile positioning GPS data continues to grow,big data-driven location-based friend recommendation systems have received widespread attention.With accurate positioning service technology,researchers can know the location information of users at any time.By analyzing the location name information that users sign in,they can further discover important information such as user interests and hobbies.In order to recommend more neighbors to the network users,so as to improve the performance of the location-based friend recommendation system,this paper proposes a friend recommendation algorithm based on spatio-temporal relationship.The algorithm mainly considers the information related to the user's specific location and check-in time.Using mathematical modeling to establish a spatial metric model of the location distance and time span in the three-dimensional space,combined with the user social relationship and the similarity of user interest and hobbies based on location mining,comprehensively consider the metric formula for evaluating the similarity of users.The algorithm is compared with other algorithms in the same gowalla dataset,and the accuracy and recall rate are better.