Web-oriented Social Network Data Acquisition And Analysis Platform

Recently, with the development of web technology and mobile Internet technology, online social networking sites represented by micro blog have become an important tool in people’s daily communication and entertainment. As an social media, online social networking sites not only has the characteristic of news propagation, but also has features of social network. Those users in the online social networking sites have formed some social networks with complex structure and large-scale data due to users’ interaction and communication. It has brought great challenges for social network data analysis and research. In order to do social network data research well, this paper has built a social network acquisition and analysis platform for online social networking sites. And this paper focuses on micro blog diffusion analysis and micro-blogging user relationship analysis in micro-blogging social network with the help of the platform. The main research work of this paper has the following three aspects:Firstly, this paper builds a social network data acquisition and analysis platform to provide the base support for large-scale social network data analysis. The overall platform uses layered software architecture to support system scalability. In the aspect of data acquisition, the platform integrates different kinds of data collection methods to collect different types of data and then stores them. In the other aspect of data analysis, for social network data, the platform encapsulates common social network data analysis algorithms and network visualization layout algorithms to provide visualization analysis function.Secondly, this paper uses the platform’s data acquisition and visualization analysis capabilities for micro-blogging diffusion visualization analysis. The platform collects two data samples to make comparative experiments, and by analyzing the characteristics of micro-blogging message in the micro-blogging social network, the platform uses visualization analysis techniques to analyze the diffusion network of micro-blogging message. After analysis, we get the statistical distribution of those characteristics and find out the great role of the key users in the process of micro-blogging diffusion.Finally, based on the platform’s processing capability of large-scale social network data, this paper also studies how the features of network structure affect the formation of micro-blogging network. The features of micro-blogging attribute are also analyzed and introduced to build a link prediction model based on random forest classifier. The link prediction model is tested on a user data set collected from the platform. By comparing the prediction performance with and without the introduction of micro-blogging attribute features and analyzing the importance distribution of features, we find that besides the network structure features, micro-blogging attribute features have significant effect on the formation of user relationship, and can improve the prediction performance significantly.