The Research And Application Of Community Detection Algorithms

In the big data, people arebecoming increasingly interested in these methodsacquiring the effective information from the vast and complex data. Complex network happens to be an important breakthrough point to study these complex systems, and the community structure is a very important feature of complex network. Anaccepted definition of community structure:the connection betweennodes within community is more closely,and the connection between communitiesisrelatively sparse. Using community detection means, the network’sinternal structure, interactions and regular features can be obtained, such as statistical properties, function characteristics and evolution law, etc. It not only provides an effective way to solving practical problems, but also greatly reduces the research’scomplexity.Therefore, the related research of network community detection attracts people’sattention.First, the thesis expounds the background and significance of this task firstly, and expatiate the research content and method to the thesis. Then, a research technical route is proposed based on community detection algorithm and network visualization technology, and the core technologiesinclude: (1)according totheseexisting community detection algorithms, putting forward a community detection algorithm based on probability method;(2)combining the community partition algorithmwiththe graph layout algorithm, it solves the problem of vast data and complex information show, and provides effective tools for network community analysis;(3)in the analysis of network evolution, the time dimension and the similarity between the communities are used as two quantitative indicators in the community evolution analysis process.In the end, the knowledge discovery research routeis applied to detecting research hotspots and network evolution analysis in domestic forestry.ln the research, the forestry literatures from the Chinese forestry information website are taken as the research object. We analysis the literature information’s complex network from the topology statistical characteristics、features and community evolution’s perspective;These forestry hotspotsare verified from the two aspects of micro and macro. The resultindicatesthat these communitiesare effective, and showsthat the above knowledge discovery research route is feasible and effective. Finally, we probe these forestry hotspots’ evolution process from the perspectives of time and similarity at different times.