Application And Contrast Research Of Centrality-based Algorithms In Complex Network Analysis

In recent years,benefit by the rapid development of data science,more and more information appears in the form of data in people's field of vision.How to mine useful information from these data has become an urgent need.Great progress has been made.Natural language processing,as "the pearl of the crown of artificial intelligence",has attracted many scholars to participate in its research.Text mining is one of the key research areas.In fact,text can be viewed as a network of different words connected to each other according to some underlying relationship.A network,also called a graph,is a unique non-European data structure that models a set of objects and their relationships.In recent years,network structures have received more and more attention due to their powerful expressive capabilities.There are many network structures in the real world.Because the topology of such networks often has high complexity,they are called complex networks.Centrality is an important concept in complex network theory.It can measure the importance of nodes and edges in the network.At present,scholars have proposed a variety of centrality-based node importance ranking and community discovery algorithms,but most of them are designed based on heuristic concepts without considering the mechanism of centrality itself and the centrality of network structure Impact.Based on the above,the main research contents of this article are as follows:(1)The design principle of the algorithm based on the centrality is studied,and the working mechanism of one of the community discovery algorithms based on the centrality is analyzed.Aiming at the inconsistency between the algorithm and the central working mechanism in network partitioning,a Improved community discovery algorithm,and comparative experiments show the effectiveness of this improved strategy;(2)Construct text networks with different topologies,compare the performance of the above-mentioned centrality-based algorithms when dealing with keyword extraction and topic recognition problems on these text networks,and through the experimental results,the influence of text network topology on centrality in dealing with different text mining problems is analyzed.Experimental results show that partial local centrality has an effect that is close to or even surpasses global centrality when dealing with keyword extraction problems.In topic recognition,local centrality often has better accuracy and computational efficiency.