Research And Application Of Community Discovery Algorithm In The Directed Social Network

With the rapid development of Internet technology,many aspects in people's life such as Entertainment,Friends-contacting,and Consumption of materials and information,are becoming more and more networked,intelligent and terminal.There are many Networking Services Platforms in people's life,such as Sina Weibo,Taobao,Sina blog,Douban.com,and Dianping.com.Among all of these internet systems,the connection among subjects can be abstracted as network diagrams and be analyzed and studied.Community discovery,as the focus of the work in social network analysis,can help to understand the internal topological structure,the functional taxonomies and evolution law of the system,and can also be applied to array of important applications,such as exploring the commercial value of platforms and improving the system's service quality.With the research on the mathematical characteristics of social network deeply,the researchers found that many networks have common characteristics: small world properties,power-law degree distribution,network transmission,community structure,community structure as an important network properties around its studies have found a large number of community algorithm.However,in the past,most community discovery algorithms were only applicable to the undirected network,ignoring the direction of the side,and therefore the result of community partition was unsatisfactory.The paper first presents research background,research significance and research status about directed social network,and also introduces the thesis framework and main work of the paper.Secondly,it illustrates the concept of directed social network in detail,including its meaning,representation,and characteristics.Besides,it introduces the community,community discovery,community structure,metrics and other related basic theoretical technology.Based on the previous researches,our paper puts forward the improved heuristic NSRC algorithm of community discovery with local searching ability,which absorbs the Six Degrees Dissemination Mode and Triadic Closure Theory,and is able to identify the initial community in the Internet.Besides,the efficient strategy of local searching,enables the candidate node implemented in the direction of increasing community modularity and brings a lower spatiotemporal complexity.The experiment on the real directed social network Data Set shows,the NSRC algorithm has better performance and the result of community partition is more reasonable and stable.Based on the PageRank algorithm applied to Web page ranking,the paper analyzes application scenarios of blog recommendation and puts forward the PeopleRank algorithm to measure the blogger's influence according to blogger's social relations.Besides,the paper combines the PeopleRank algorithm with the NSRC algorithm and applies them to Blog recommendation on the Blog Social Network.With these algorithms,the blog finally realizes the system of blog recommendation,which simulates the real Blog Social Network Services platform,and achieves a blog recommendation system with functions of user management,blog management,and articles recommendation in one.