• •

### VisIOT:经济产业结构关联可视分析

• 出版日期:2019-11-25 发布日期:2019-11-28

### VisIOT: Visual Analysis of Economic Structure Correlation

Zhou Zhiguang Shi Chen Hu Miaoxin

• Online:2019-11-25 Published:2019-11-28

Abstract: With the refinement of the category of economic statistics and the standardization of statistical process, the input-output table (IOT) data that can effectively measure the correlation of industries are increasingly presented with complex structural characteristics. The traditional statistical analysis software and method can only convey limited information through a few simple forms. Faced with the complex and dynamic evolution of IOT data, it is difficult for users to effectively analyze the hidden relationship and explore the characteristics of the temporal sequence. Therefore, this paper designs a visual tool named VisIOT to explore the related features in economic structure. First, Two-way force-directed graph is designed to describe the associated network topology structure in the economy. We then use matrix diagrams to visually show the differences between IOT data, and embed the multiple data into the cell in a temporal sequence order, building the chronological matrix diagram. Furthermore, we construct the chronological force-directed diagram, using the temporal and spatial attributes of IOT data to map the network nodes and connections. To effectively analyze and extract the industrial groups in economic relation structure, the modularity algorithm is used to dig into the connotative characteristics of community structure in the association relationship. After that we design a temporal community evolution diagram to show the temporal evolution pattern of community structure characteristics, which helps users effectively capture the community structure stability. Finally, convenient interactions are also integrated in this system for a visual interface to realize the visual analysis system of economic structure correlation. We use real IOT data for case analysis and verification, and the results show that our method can help users to quickly identify and perceive the associated characteristics and temporal variation patterns implicated in IOT data.