Research On Information Mining Based On Investment Graph

With the development of society and economy,my country's equity investment market has grown rapidly.The equity investment market provides financing for thousands of enterprises,thereby supporting the development of enterprises and promoting the sustained growth of the Chinese economy.However,equity investment also brings many uncertainties.The specific manifestation is that the investment in the equity market makes the connection between enterprises closer.Once some important entities in the financial market have risks,they may affect the entities closely connected with them.Which eventually evolved into systemic financial risks.The financial system is an important part of a complex system.If information about important entities and communities in the financial market can be unearthed,the supervision of important entities and communities in the financial market can be strengthened to effectively prevent systemic finance risk,maintaining the stability of the financial market,and promoting the development of complex system research,can also reflect the characteristics and advantages of the combination of theory and practice and the intersection of disciplines possessed by the control engineering specialty and promote the development of control engineering discipline.Therefore,how to effectively dig out important entity information,community information and investment relationship information between entities in the financial market has become a current research hotspot.The research in this thesis is mainly carried out from the following three aspects: identifying important entities in the investment graph of listed companies;dividing the entity community relationship with important entities as the core in the investment graph of listed companies;mining the investment relationship between two entities and a single entity Investment structure.The main work of this thsis is as follows:Firstly,based on the investment relationship,an investment graph between companies is established,and an improved Page Rank algorithm is proposed to identify important entities in the investment graph.In order to improve the Page Rank algorithm,the traditional Page Rank algorithm only considers the shortcomings of in-degree,and fuses the ratio of the entity outdegree to the total number of links based on the traditional PR value.The experimental results show that compared with the traditional Page Rank algorithm,the improved Page Rank algorithm has a better effect on the distinction between entities and the importance of entities.Secondly,based on the above important entity recognition algorithm,a community division model based on Page Rank-Deep Walk-Cosine model is proposed.From the perspective of information flow in the investment graph,the improved Page Rank algorithm is used to first identify important entities in the investment graph,and then the Deep Walk model is used to vectorize the entities in the investment graph,and the cosine distance formula is used to calculate the vector between the vectors.The degree of similarity is based on the degree of similarity as an indicator for dividing community relations.Experimental results show that the model can well divide the community with important entities as the core.Thirdly,based on the investment graph of listed companies,the investment path information between entities and the investment structure information of a single entity are mined.Find the investment relationship path between the two entities through a backtracking algorithm,deeply analyze the investment relationship between the two entities,and reveal the internal logic of asset price linkage between companies that are not directly related;use the breadth-first algorithm to carry out equity.Thoroughly explore the actual controlling party behind the entity and reveal the logic behind asset price linkage.