Study On Items Recommendation System Based On Heterogeneous Information Network Analysis

With the development of the Internet, e-commerce and mobile Internet, a growing number of items trading are in electronic form. The transaction of the Internet electronic items trading platform increase and grow cumulative. Faced with the growing e-commerce transaction data and user requirements, how to quickly and accurately to recommend the goods of meeting the needs of users is a key issue of the current scientific and technological development in the field of e-commerce. In the paper, the National Natural Science Foundation of China, "Web service recommendation based on heterogeneous information network analysis" as the support foundation, and heterogeneous service network analysis as the reference, items recommended based heterogeneous information network analysis is proposed.Based on the recommendation method, clustering analysis and information network research, it construct heterogeneous items network model which combined with the requirements of the e-commerce platform and considering the potential associations of various categories participating in e-commerce transactions; Next is the network clustering and sorting based on transaction data analysis, combined with a certain recommendation strategies to achieve the items recommendation. From the relationship of buyers, sellers, items, terms, to precede heterogeneous items network, it considere to the relationship of the recommendation between the four items trading process and improve the efficiency of items recommendation.The details of research works in this paper include:① Combining the trend of the heterogeneous information network development, research status and analysis the problems based on network analysis, heterogeneous network analysis and recommend merchandise and other domestic and proposed research and innovation point.② Studies describing heterogeneous information network model with formal methods, combined with characteristics of e-commerce transaction data structure and maintaining of heterogeneous items network.③ Proposed items clustering based on heterogeneous items network analysis with heterogeneous networks description model. Analysis on heterogeneous items clustering with the Ranking function、Ranking model and Net Clus algorithm.④ From the dimensions of the relationship between items trading record, with the clustering results, combined with the corresponding recommended strategy recommended electronic goods, items were recommended by a detailed description of the recommended ideas, recommended processes.⑤ Based on heterogeneous items network clustering result, proceed from the idea of software engineering, design and implement the prototype of items recommendation. Describe the prototype of the design process from requirements analysis, system design and implementation effectiveness analysis.