The Construction And Analysis Of Enterprise Relationship Network For Precursor Chemicals Industry

Precursor chemicals are the industrial production and also the materials to make drugs.So it has great significance to make effective supervision of transactions among enterprises in precursor chemicals. In order to meet the regulatory requirements of making decision and prediction of trend, we use the theory and methods of complex network to make an enterprise as a single node and the relationship between the two enterprises as an edge of the enterprise relationship network. Based on these, we can study the relationship among enterprises and the transaction of precursor chemicals. At present, the research on the precursor chemicals industry relationship network is rarely made. The main work of this thesis is as follows:1. Data extraction and arrangementThe data constructed the enterprise relationship network is from The Information System On Supervision And Management Of Precursor Chemicals, which consisting of the precursor chemicals trading data of Hebei Province, Hunan Province, Hubei Province, Liaoning Province, Fujian Province, Guizhou Province, Shanxi Province, Anhui Province and Tianjin city. Based on the understanding of the precursor chemicals industry’business, we analyze the structure and data of the data source. Thus we extract the data needed to construct the enterprise relationship network, and clean and tidy the noise data and historical data in the data.2. Construction and graphic exhibition of enterprise relationship networkThe enterprise relationship network is a directed and weighted network and its node represents a single enterprise, including the sales enterprises and the purchase enterprises.And the directed edges between nodes are the enterprise relationship including purchasing relationship and transporting relationship(this thesis focuses on the purchase relationship). In addition, the constructed network is exhibited in graphic effect.3. The analysis of the enterprise relationship networkBased on the understanding of the practical characteristics of the network and the complexity of the enterprise relationship network, we divide the enterprise relationshipnetwork into two networks, including directed and unoriented enterprise relationship networks. By analyzing the statistical characteristics of enterprise relationship network in different perspectives, we find that the network has similar characteristics that greater clustering coefficient, smaller average path length, exponential degree distribution and so on with the existing social network, and it also appears the enterprise relationship network is negatively correlated by the degree with the social networks.The result of this thesis extends the basic research of complex network and provides a scientific basis for the fine analysis of transaction behavior in the application.