Research And Optimization Of Socail Network Matching

Social activity using a variety of online social networks has become a feature of the current social. Each online social network reflects one aspect of our true social network. The social network matching algorithm that connects multiple online social networks plays an important role in our overall social network analysis. However, the current online social network matching algorithm has high time complexity and low matching accuracy in the case of less information between graphs. It has no good combination with other domain knowledge to improve the practical application ability.This thesis studies the time complexity, matching accuracy and application of social network matching algorithm in practice. This thesis proposes a social network matching algorithm based on subgraph segmentation, a social network matching algorithm based on directed graph, and a friend recommendation algorithm based on multiple social network matching. And in the experiment achieved very good results.The main results are as follows:1. For the time complexity of the social network matching algorithm is high, considering the high time complexity of the algorithm is mainly due to the existence of a large number of algorithms in the calculation of redundancy, we can reconstruct a number of subgraph.And the community segmentation algorithm is introduced in the process of refactoring, so that the relation between subgraphs is relatively sparse,and the connection between subgraphs is relatively close. The social network matching algorithm based on subgraph segmentation can save the time of matching in the process of social graph matching, and save 30% time by reconstructing the subgraph and index searching.2. The algorithm of social network matching based on directed graphs effectively solves the problem that the matching accuracy of directed graph is not high,and proposes a direction-based node similarity calculation method. By computing node similarity of two points in the network layer The directionality is introduced and the accuracy of matching algorithm is improved. The accuracy of the algorithm can be increased by 10% when the social graphs are matched between the social network graphs which are sensitive to the directional change.3. The recommendation algorithm based on multiple social network matches the traditional friend recommendation algorithm and social network matching algorithm to improve the recommendation accuracy.