Research On Mining Algorithm Of Social Networks Node Based On Greedy Sub-graph

With the development of computer technology such as big data and social computing,the research of social networks has made a great strides forward.The scale and completely of social networks are constantly increasing,and today's computers have the ability to calculate and analyze them.Enterprises promoting their products need to target its key customers;educational institutions training courses need to find its core audience;government guidance of public opinion need to find the source information.With the increasing amount of information,when large-scale social networks and complex social networks appears,it is important to be able to effectively excavate the key nodes in the network,and this has become an important part of social networks research.The method of node mining in social networks is divided into heuristic method and greedy method.The former mainly measures the importance degree of each node in the network according to its own attributes or network topology,such as the centricity algorithm,only the neighbor topology of the node is taken into account,although it is fast,the accuracy is not good;another example is closeness centricity algorithm and betweenness centricity,the calculation involves the whole network,so the efficiency is low.While the latter propagation simulation is through the propagation model for each node,and then calculate the importance of the node through the size of its propagation,this kind of algorithm is inefficiency for the real spread is combined with the propagation model,resulting in its not applicable in large-scale social networks.Based on the bad node mining results in heuristic method and the highly complexity in greedy algorithm,we take the heuristic method first and then use the greedy method,proposes a new node mining method based on greedy sub-graph algorithm in this paper.Firstly,calculate the impact potential of the node according to the important attribute of node degree and the local topological structure,then add the node to seed node candidate set according to the sorted impact,at the same time,the seed node candidate set is selected by ranking the whole network and choosing the node with the highest specific threshold.After the selection of the candidate set,the linear threshold model of the influence is represented as the greedy sub-graph strategy.The node with the largest incremental impact is selected to join the finalnode mining result set,and the nodes in the candidate set are dynamically modified at the completion of each step of propagation,the candidate set correction process is repeated and the simulation process is propagated until the set of node mining results of the expected size is reached,and finally an ideal node mining effect is obtained.Then,comparing the heuristic algorithm,the greedy algorithm and the algorithm proposed in this paper on the effect of node selection and the effectiveness of algorithm and the spread range,the node mining algorithm based on greedy sub-graph can guarantee good optimal solution on the basis of ensuring the node selection effect and efficiency better than the classical algorithm,and verify the theoretical applicability and actual propagation effect of the proposed algorithm.