Research And Implementation Of High-quality Enterprise Discovery Algorithm Based On Blockchain Industry

As an emerging technology,block chain is tamper-proof,traceable,decentralized.What is more,it can be applied to supply chain management,credit management,product traceability and other aspects,so it has attracted much attention.Since 2019,blockchain industry has been booming in China.Guidance and support policies on blockchain development have been issued from the central to local authorities.According to statistics,in the first half of 2019,more than 23 provinces and cities in China released 112 policy information related to blockchain.Many provinces and cities have incorporated blockchain into their plans to develop the digital economy.With the gradual improvement and development of blockchain technology,more and more enterprises are engaged in the research and development or investment of blockchain.The government is also paying more and more attention to the development of blockchain technology.At present,blockchain project has become one of the investment hotspots.However,the blockchain projects and related enterprises in the market are mixed with good and bad.It is difficult to select worthy companies with high quality.so it is of great significance to select the ones worthy of investment from the numerous enterprises.Most enterprises in the blockchain industry are unlisted.They are small and medium-sized emerging enterprises.Their financial data required by the traditional enterprise value evaluation model cannot be obtained.Therefore,this paper tries to find the important enterprise nodes in the blockchain industry from the perspective of complex network.Network important node identification has always been a hot research topic.There are many methods to identify important nodes such as degree centrality,intermediate number centrality and other node attributes.Besides,there are k-kernel decomposition,Page Rank algorithm,Leader Rank algorithm based on random walk,etc.Through comparison and analysis of various algorithms,this paper determines to use the Leader Rank algorithm.This paper considers the attributes of enterprise nodes in the iteration process to identify and discover important enterprises in the block chain industry.In this paper,a total of 681 blockchain enterprises are as research objects,including listed by Shanghai science and technology commission and listed by cyberspace administration of China.The basic information,investment and shareholder information of these 681 enterprises were obtained from tianyancha,a website containing information of many businesses.These data were analyzed to construct the evaluation index system of enterprise value competitiveness and the cross-shareholding network in the block chain industry.According to the overseas investment and the relationship between shareholders,the cross-shareholding network is constructed,in which the proportion of investment is the weight of the edge in the network.The evaluation index system of enterprise value competitiveness is constructed from three dimensions of enterprise scale,capital chain and influence in the network constructed by presidents of every companies.Entropy weight method is used to determine the weight of each index and score each enterprise.In the iteration process of the Leader Rank algorithm,the enterprise's value competitiveness score is used as a part of the node weight,and different scores are assigned according to the different value scores of different nodes in each iteration.When verifying the algorithm results,considering the particularity of the cross-shareholding network of enterprises,this paper does not use the commonly applied SIR model,but compares the results of this paper with the enterprise rankings published by other authoritative institutions.By comparing the algorithm in this paper with Page Rank algorithm,Leader Rank algorithm and weighted Leader Rank algorithm,it can be seen that the algorithm in this paper has higher accuracy.Therefore,the heterogeneity of nodes is fully considered in the identification of important nodes in the network,and the algorithm results will be more accurate by combining the attributes of nodes.