Recommendation Method Based On Knowledge Graph In Science And Technology Service Industry

As an emerging industry,the science and technology service industry has now played a very important role in the modern service industry.With the explosive growth of scientific and technological achievements,how to quickly find and recommend high-quality matching technology service resources for technology practitioners or researchers has become an increasingly important subject.There are many types of science and technology service industry,among which scientific research and technology consulting services account for a large proportion in the science and technology service industry.How to recommend scientific and technical literatures of interest to scientific research staff and how to provide expert consulting services for scientific and technical practitioners are the focus of this article.Traditional recommendation systems mostly recommend based on content or collaborative filtering,which often has shortcomings such as cold start and poor interpretability of recommendation results.In order to solve the above problems,this paper constructs a scientific and technological service knowledge graph by fusing multiple data sources,mainly involving research and development,scientific and technological consulting and other scientific and technological service fields.Based on the information provided in the knowledge map of scientific and technological services,when recommending scientific papers to scientific researchers,they can use their published paper information,personal information and community information to provide a priori information to alleviate the cold start problem.This paper proposes a method of scientific and technological literature recommendation combining user preferences and knowledge graph path information.The user preferences are constructed through the paper keywords in the knowledge map of scientific and technological services,and user operation information,and the knowledge map path information is combined to recommend scientific literature for users.Due to the large number of nodes in the knowledge graph of scientific and technological services,generating paths in the general BFS way will consume a lot of computing resources.This paper proposes a two-way BFS method based on delay expansion.This method can quickly generate the path between two nodes.Use LSTM to extract information of multiple paths,and combine the user preferences to recommend papers of interest to users.Science and technology practitioners may consult relevant experts on professional issues during their work.In this paper,we use the FP-growth algorithm to mine the frequent subsets of expert Q&A records in scientific and technological consultation,and integrate the extracted expert co-occurrence information into the scientific and technological service knowledge graph.Then use Deep Walk to extract graph features and generate expert and topic node word vectors.In this paper,an improved Deep FM model is proposed based on the extracted graph features,statistical features,text features and time features of the problem.Use this model to provide scientific and technical personnel with recommendations for scientific and technological consulting services.