Interactive Visual Analysis Of Internet Data

With the wide application of the Internet, more and more people are involved in Internet events and Internet based social network. The Internet has been one of major sources of information. It gets easier to grab data for Internet based research. Research on Internet data has been a new field of research since application of the Internet. It has the following characteristics:liberty, interactive, diversity, variation and sudden. People are willing to share their information on the Internet and expand their contacts. People could release their relevant events quickly and timely using Sina microblog or twitter. These events could the those they are undergoing. So social media complement for the short-comings of traditional news media. Enterprises could analyze and mine the huge amount of network data and identify great potential business opportunities. Companies could even affect customers directly through the social platform. At the same time, customers also can get information from the network to understand all aspects of the enterprise. And they could achieve the investment objective guidance and decisions based on these data.Information visualization aims to study visual representations of abstract data to re-inforce human cognition. It could help to understand and analyze information by showing the abstract characteristics of data using graphical methods. Diversification of the Internet data causes huge challenges and makes the analysis of traditional statistical analysis and mathematical analysis incapable of action. Therefore, the Internet data visualization has very important significance.Firstly in this paper, time-varying analysis and space-varying analysis are combined together, and we proposed the Electron Cloud Model (ECM) which is based on the Schrodinger equation and Niels Bohr atomic theory. Electron Cloud Model is used to per-form time-varying visual analysis of micro-blog data. In the ECM, we tried to mapping a score of sentiment to the electron stability. Kernel density estimation and edge bundling were applied to conduct space-varying visual analysis of sentiment. Kernel density esti-mation visualizes data changes in different levels of detail naturally while edge bundling is used to reduce visual clutter of edge crossing and reveal high-level edge pattern.Then, we conducted visual analysis for collected enterprise data. We implemented an enterprise knowledge graph system, and analyzes the problems of enterprise knowledge graph visualization encountered in the process. We applied group factor and Voronoi diagram to the force-directed algorithm. They respectively solved too many intersection caused by set(or group) edges and overlapping between set edges and elements.