Discovery Of Online Review Theme Community Based On Complex Network

With the rapid development of the Internet,the number of online users is also rapidly increasing.This has led to exponential growth of data on the Internet,especially ecommerce platform review data.Enterprises and consumers have been unable to meet the need for manual screening and information acquisition.Online reviews are used for quick topic mining to make effective community divisions for users.Based on the online review topic community mining,it can provide decision-making information for consumers’purchase behavior,and also provide technical support for merchants to improve their products and accurately target the consumers.The traditional online comment mining method is often based on statistical methods,ignoring the semantic relevance information between online reviews,resulting in problems such as low mining accuracy and low processing efficiency.In recent years,the research on the theory and practice of complex networks has become more and more mature.The language has also been proved to have the characteristics of "small world".More and more scholars have begun to apply complex network theory to natural language processing.In order to solve the problem of the lack of semantic association information in the traditional online comment mining method,this paper introduced the complex network theory into the research of online review topic community discovery.The main content of this paper is to build an online commentary network and a topic community discovery for the network,the main research content in this thesis is as follows:Firstly,this paper systematically expounded the theory of complex networks and their applications in language processing,the theory of text representation and feature selection,and the theory of community discovery algorithms.This paper constructed a single online comment complex network based on complex network text representation model and simplified it based on complex network feature selection algorithm.It further combined the maximum common subgraph related theory to realize the similarity calculation between online comments at the semantic level.Based on this,the specific construction process of the online review network was proposed.Secondly,the paper analyzed the community structure characteristics of online comment network based on the theory of community structure,and proposed the LFMS algorithm to optimize the LFM community discovery algorithm based on the structural characteristics of the online comment network and the semantic similarity between nodes.Finally,the paper combed the evaluation indicators of the community discovery algorithm,and designed two groups of experiments to verify the rationality of the online review network and the effectiveness of the LFMS algorithm against the actual review data crawled by the network.By analyzing the experimental data,it was concluded that constructing an online commentary network based on complex network theory had certain rationality.The improved LFMS algorithm could improve the effect of online commentary topic community discovery to some extent.