Microblog cell division method based on user comprehensive similarities

Abstract

The invention designs a microblog cell division method based on user comprehensive similarities. According to the specific process of the method, 1, microblog data is acquired, LDA topic model training is performed on a blog article set, and a user topic similarity matrix is obtained through topic mining based on feature extension; 2, a network topological graph with users being nodes and user relations being edges is constructed, and a user comprehensive similarity matrix is obtained according to node link relevancy and topic similarities; and 3, a unique tag is allocated for each node first,the potential influence of each node is evaluated, then the descending order of the potential influences serves as a node selection order, the descending order of node comprehensive similarities serves as a tag update order of the nodes, and finally iterative update of the tags is performed. In this way, cell division can be performed on the microblog users through an improved tag propagation algorithm on the basis of considering the user comprehensive similarities, and the method has high application value for online public opinion monitoring, commercial user mining and the like.

Description

technical field [0001] The invention relates to the technical field of social networks, in particular to a method for dividing microblog communities based on user comprehensive similarity Background technique [0002] How to mine useful information from social networks has become a research hotspot in complex networks, which is of great significance both in theory and in social practical value. Because many social networks have a community structure, especially in large-scale social networks, a common feature of this network structure is that the connections between nodes in the community are very close, and the connections between communities are very sparse. As more and more scholars invest in the research of community division, it is found that the algorithm based on the community structure is characterized by greatly reducing the scale of network processing, which greatly improves the time performance efficiency and avoids the influence of nodes to a certain extent. The...

AI Technical Summary

Problems solved by technology

In 2007, Raghavan et al. proposed the Label propagation algorithm (Label propagation Algorithm, LPA), which effectively solved the problem of high complexity and unconvergence.

[0005] To sum up, there is a lot of room for improvement in the accuracy and stability of the existing community division methods.

Images