Research And Realization Of Multi-dimensional Evaluation Of Social Network Influence Maximization

With the continuous improvement of the Internet and the rapid development of network technology,social networking applications have become a part of People's Daily life.With the increase of the amount of data,social network analysis of large data into a challenging research area,the amount of data contain the user generated content and complex relationships,makes popular influence maximization problem,is one of the direction of the main problem is how to quantify the impact of each user,and how to identify the most influential social network users.Based on douban city activity data,this paper aims to propose an impact maximization algorithm with wider applicability and more comprehensive factors.First,we study the issue of impact maximization based on time-subject.From users to participate in the activities of the theme of historical information can get the user distribution,so that users to participate in activities can reflect the theme of the user preferences,and participate in the activities of the more recent can reflect the theme of the user preferences,so this article will time and subject factors applied to the PageRank algorithm,the influence of time-topic model is put forward,the results show that the proposed algorithm is more effective to assess the effects of a user.Secondly,location-based impact query is studied.Since the activity is held offline,only studying the user's theme preference cannot meet the demand,so the user's location preference should be considered.An lr-tree index structure is designed,in which each node stores the user's theme and location preference,which can effectively identify the seed node for a given query request.In order to improve the efficiency of seed selection,the approximate algorithm and heuristic algorithm based on lr-tree are proposed.Finally,we design and implement an impact maximization query system based on the algorithm in this paper.