Research And Implementation Of Influence User Discovery System Based On Network Representation

With the rapid development of Internet technology,online social networking has gradually become a way of daily life for people.However,with the acceleration of the accumulation of social data,it is difficult for people to quickly find the desired data resources and respond to relevant information in the context of massive data.Therefore,designing and implementing a system for discovering influential users in social networks has great application value.This paper divides the network community on the results of the network representation algorithm.Then,the research on maximizing network influence in subnets is carried out.The main contents of this paper are as follows.In terms of network representation,a network partition method based on network representation is proposed for the large amount of network data and the uneven distribution of data.This method uses an arc-based dynamic self-learning network representation algorithm.The measure of similarity between nodes is similar to the propagation process of similarity.Neighbor nodes with high similarity to the target node are selected,and the nodes in the network are represented by Skip-Gram method.Experimental results show that the algorithm proposed in this paper has a maximum increase of 22.62%in the F1 value compared with the comparative experiments in the node classification task.In terms of influence maximization,in order to solve the problems of high computational complexity caused by the hyperparameters in algorithms such as PageRank and ignore the influence of user behavior factors,this paper proposes a method of global influence ranking.This method improves the LeaderRank algorithm from two aspects:regular transfer probability and random transfer probability.Not only does it take advantage of Ground Node to make the network fully structurally connected,but also increases the semantic interpretability of user behavior.The experimental results show that when we use the independent cascade model for evaluation,the influential Top-k users selected by the algorithm in this paper can affect more users in the comparison experiments.Design and implement an influential user discovery system based on network representation community division.The system adopts a modular design concept and mainly includes four modules:data processing module,network presentation module,community computing module and system management module.The system can find the community with the least modularity given the network data,and then find the influential users in each community.The system is easy to operate and response quickly.