Research On Visual Analysis Methods On Associations For Hierarchical Data

The exceeding of pesticide residue is one of the most important factors that lead to the food safety problems. In order to use pesticides rationally and efficiently and to supervise the exceeding use of pesticides, the government detects the pesticide residue situation of all kinds of agricultural products, where the government can get abundant pesticide residue data. Most of these data is of hierarchical structure. It is a problem demanding prompt solution to analyze the hidden association relationships in the pesticide residue information aiming at these data. Based on the research on the pesticide residue detection data, with the combination with the techniques of information visualization and visual analysis, this paper studies on the association relationship of hierarchical data, and contributions of this paper are as follows:Firstly, sunburst with ordered nodes based on hierarchical clustering is presented.Expressing not only the hierarchical structure but also the association relationships hidden in the hierarchical datasets is a problem not only important but also challenging. The current methods express association by connected lines, which can cause visual clutters and is difficult to achieve ideal effects on big data. In this paper, hierarchical clustering and sunburst are combined, and the sunburst nodes are sorted. The sorted sunburst expresses the hierarchy with radial layout and expresses the association relationship with the locations of the nodes. The algorithm is applied into the pesticide residue detection data,and user studies are well done. The results show that the algorithm not only improves the utility rate of space, but also offers convenience for users to inquiry the association relationships between pesticides and pesticides.Secondly, a hybrid layout algorithm for double interrelated tree is proposed. Aiming at the needs to analyze the association relationships between two kinds of hierarchical data,hybrid layout is used and node-link and sunburst are combined. Bézier curve is used to optimize the layout and reduce the visual clutters by replacing the straight line edge by relationship edge, which can make the distribution trend of the association relationships much more obvious. The optimized double interrelated tree was applied into the pesticide residue detection dataset. The layout result can clearly show regional hierarchical information, pesticides’ classification information, and relationships between them.Thirdly, an associated visual analyzing system for pesticide residues detection results data is designed and implemented. This system implemented the collection of the nationalpesticide residue detection data and the construction of its database, combining sunburst with ordered nodes based on hierarchical clustering and hybrid layout algorithm for double interrelated tree, having analyzed and processed the information of regions information and agricultural products and pesticides detection results information in the pesticides residue detection data, which is visualized by multiple views. The purpose is to search and analyze better the association relationship of hierarchical data.The work of this paper takes “The Twelfth Five-Year” National Science and Technology Program “Collection, management, intelligence analysis research and system implementation of pesticide residue detection data in food” as background, combining visualization methods and data mining techniques to achieve the task of visual analysis on associations for the pesticide detection data, which can offer technical support for decisions and supervisions. The techniques that this paper offers also can be applied to associated visual analysis of other fields like the financial Information, business information, network security, social network et al.