Weibo Community Detection Based On Interest

As social media flourish, hundreds of millions of Internet users around the world exchange knowledges, share ideas, post messages and do a lot of daily activities on the Internet, so that the Internet has become an area of vibrant and energetic. These vast amounts of information is collected, collated and published by each society individual, while countless users to read and disseminate and give timely feedback. It is,to a certain extent, represents an online version of human society. It has accumulated vast amounts of information that can be stored, searched and related to all aspects of human life, contains a huge value.Weibo social media network is an outstanding representative in China, which is rapidly development since the establishment of micro-Bo. Today, it has shaped Chinese society and has a huge impact on our life. A huge amount of Weibo data has been accumulated, we need to study it and digging through Weibo research data. In this way, we can not only have a better knowledge and understanding of human behavior patterns, but also can better guide the product and thus better serve our life.Indeed,social network is a network consisting of nodes and edges. Community is a network of sub-graph nodes within the community as closely as possible, the junction between communities as sparse. Social network community understanding and discovery is an important way to understand social networks.Current scholars have raised a lot of community discovery method. But specific to the Weibo network, they still need to be improved and optimized. According to the characteristics of Weibo network, this article focus on Weibo community network discovery method. The main research work is as follows:(1) Based on analysis for customer relationship in Weibo network, according to their specific link properties between users, a new user similarity and community similarity measure are raised. Considering the idea of community-level, this article try to build a suitable Weibo network community discovery method and shows its effectiveness by experimental analysis.(2) To make full use of Weibo network information, we try to make Weibo content extracted and quantified. Then,combining upper link similarity algorithm, we propose a framework for integration of content and links D-CODICIL. Lastly, through analysis, we demonstrate its effectiveness.(3) In order to adapt to the Weibo network of massive data, to solve content and links algorithm relatively time-consuming phase, this paper attempts to transform it into distributed algorithm. Based on Hadoop’s MapReduce computation model, the paper propose three MapReduce jobs and make it parallelization. The results show that Hadoop applications can greatly improve the computational efficiency of the algorithm.