Community construction method and community construction device

Abstract

The invention provides a community construction method and a community construction device. According to the invention, an initial influence value of each node is calculated according to the attributevalue corresponding to each node in an initial community network, wherein the attribute value comprise the number of user questioning times, the number of question answering times, the number of praising times and the number of praise obtaining times; based on the initial influence value of each node and the pagerank algorithm, calculating the actual influence value of each node; selecting a nodelargest in actual influence value among all nodes of the initial community network as a core node to form a first community with one node, and marking the core node; judging whether the adjacent nodes of the first community exist in the initial community network or not; if not, completing the construction of the first community. According to the invention, a network community is constructed basedon the core node of the network and the attribute factors of each node. Therefore, the problem that the core node is fuzzy to select is effectively solved. The accuracy of community construction is improved.

Description

technical field [0001] The present invention relates to the technical field of data mining, in particular to a community building method and device. Background technique [0002] Existing community building algorithms usually include community building algorithms based on graph division such as Kernighan-Lin algorithm and spectral bisection method based on Laplace graph eigenvalues, community discovery algorithms based on hierarchical clustering such as GN algorithm, and real community tracking based on real communities. A community-building algorithm for virtual community evolution techniques. [0003] Among them, in the community discovery algorithm based on graph partitioning, the Kernighan-Lin algorithm needs to know the size of the network community in advance, otherwise the calculation error will be large, and the spectral bisection method calculates the eigenvector and eigenvalue, and the calculation amount is relatively large, which is not suitable for analysis. lar...

AI Technical Summary

Problems solved by technology

In the community discovery algorithm based on hierarchical clustering technology, such as the GN algorithm, although the accuracy of the algorithm is high, the calculation amount is very large due to the need to continuously calculate the edge betweenness of each edge of the community

In the community discovery algorithm based on dynamic network community evolution, such as the technology proposed by Hopcroft and Khan to use real communities to track the evolution of virtual communities, some parameters must be set, and it is difficult to adapt to the changing community evolution situation, which brings serious problems to practical applications. very difficult

Images