Social Network User Recommendation Method Based on Dynamic Influence

The rapid development and wide application of Online Social Network (OSN) has produced a large amount of social data. How to effectively use these data to recommend interesting relationships to users is a hot topic in social network mining. At present, the user relationship recommendation algorithm relies on similarity, and the user’s influence is insufficiently considered. Aiming at this problem, this paper proposes a new user influence evaluation model, and on this basis, a new user relationship recommendation algorithm (SIPMF) is proposed by combining similarity and dynamic influence. 2522366 Sina Weibo data were crawled to build an experimental data set for experiment. Compared with the typical relational recommendation algorithms SoRec, PMF, and FOF, the SIPMF algorithm improved 4.9%, 7.9%, and 10.3% in accuracy and recall respectively. And 2.6%, 4.2%, 6.6%, can recommend for users more interested in the relationship.