Research On Social Network Public Opinion Analysis Based On GN Algorithm

At present,the analysis of social network public opinion has attracted much attention,and the technology of network public opinion analysis is also diversified.The commonly used social network discovery clustering method is Girvan-Newman Algorithm.The advantage of the GN algorithm is that the accuracy of the community is high,the algorithm structure is simple,and the robustness is strong.However,the GN algorithm still has problems.The time complexity of the algorithm in finding the community structure is high and the calculation efficiency is slow.To solve this problem,this paper proposes the parallelization and improvement of the existing GN algorithm,so as to improve the speed of community discovery during the social network public opinion analysis,so that the community can be quickly and accurately discovered in practical applications and the community structure can be determined.This paper first analyzes the calculation process of the traditional GN algorithm,and finds that the algorithm is performed serially when calculating the edge-predicted values.At the same time,the algorithm only computes the nodes and edges generated by the breadth traversal map of one node in the network during the cycle.Relationship structure.This structure provides the a priori condition for the next edge value recorded when scanning the shortest path between points.By combining the research of parallel computing theory with GN algorithm,the parallelism of GN algorithm is obtained.Secondly,the parallel computing model of the algorithm is established,the relationships among the nodes in the sensation network are described,the connections between the nodes and the edges are analyzed,and a parallelized GN algorithm is designed based on this model.Once again,the Hadoop platform is introduced to combine the MapReduce parallelism and algorithm under Hadoop to calculate each node in the network under the multi-machine model,thereby reducing the time complexity of the algorithm and not reducing the accuracy of the algorithm.Under the premise of improving the speed of community discovery in the public opinion network.Finally,the parallelization effect is analyzed by the experimental results and compared with the results calculated by the traditional GN algorithm.The experimental results show that the accuracy of the parallel computing algorithm is the same as the serial computing result,and the discovery speed of the sensational network community is accelerated without reducing the accuracy of the algorithm.In the end,the time complexity of the improved algorithm is significantly reduced,and can be run in an engineering instance.This facilitates the monitoring of public opinion in the social network,and provides assistance for social harmony and stability.