Research And Application Of Community Discovery Method Based On Tag Propagation

In recent years,with the continuous change and development of the Internet and communication technology,various complex networks exist in real life,and related research on complex networks has also received great attention from scholars in various fields.Community structure,as an important feature that is commonly found in complex networks,is of great significance to further explore the structure and nature of complex networks.Detecting communities in complex networks can mine potential groups in the network and reveal the underlying laws or associations of groups behind the real network.It has high practical value in modern business applications and has been widely used in recommendation and retrieval fields.At present,there are many related algorithms discovered by the community,but such algorithms generally have defects such as high time complexity and low accuracy.As a community discovery algorithm,label propagation algorithm has low time complexity and does not require any prior conditions.It is suitable for community discovery in large-scale networks and has received widespread attention.However,the label propagation algorithm has the defects of high randomness and low stability.In view of the above-mentioned shortcomings,this paper improves the existing label propagation algorithm,and improves the stability and accuracy of the label propagation algorithm to a certain extent.(1)Aiming at the problem that the original algorithm did not consider the difference of nodes during node initialization,this paper proposes a label propagation algorithm based on core node layer-by-layer expansion.Optimize from node initialization and give a new label propagation method.First,the Leader Rank algorithm is introduced to calculate the influence of the nodes,and secondly,based on the influence of the nodes,the core nodes are selected as the source of initial label propagation to reduce unnecessary judgment overhead in label propagation.Finally,the node labels are updated in a layer-by-layer manner.The experimental results show that although the algorithm has high time complexity,it can detect community accurately and stably.(2)Aiming at the problem of the original algorithm's node update order and the randomness of node labels,this paper proposes a label propagation algorithm based on node influence.The k-shell algorithm is introduced.Based on this,a new influence calculation method is given in combination with the degree of the nodes.The update order of the nodesis fixed according to the influence of the nodes,and the "backflow" of resources is avoided.When the node label is updated,the local influence of the neighbor node to be updated is considered,so that the selected label is more accurate.Experimental results show that the algorithm can detect communities stably and efficiently,and the time efficiency is close to the original algorithm,which is suitable for large-scale networks.Based on the above algorithm,this paper designs and implements a microblog user interest community division system.The two main functions of this system are user influence analysis and user interest community division.By crawling Weibo user data for experimental tests,the experimental results show that the system can accurately divide user interest communities.