Research On Intelligent Optimization Model Of Power Consulting Experts Based On Knowledge Graph

Scientific and technological innovation is the source of power for an enterprise to ensure its core competitiveness.The State Grid Corporation of China,as a power enterprise in China,has gradually turned its attention to the field of scientific and technological innovation.In recent years,with the increasing investment in science and technology projects of State Grid Corporation,the research topics and science and technology projects are also increasing.Along with it,the scope of scientific research is more and more extensive,interdisciplinary science and technology projects are more and more common,and the combination of emerging disciplines and power grid is more and more close.These characteristics put forward higher requirements for the power consulting experts who participate in the approval and acceptance activities of science and technology projects.On the other hand,with the increasing scale of the State Grid expert database,it is a great challenge for the State Grid project managers how to efficiently and accurately recommend the evaluation experts in this field to participate in all aspects of the life cycle of scientific research and innovation projects.At present,the traditional manual selection of experts is still subjective and one-sided.Therefore,how to select suitable experts under the background of complex interdisciplinary and huge expert database has become an urgent problem.Under the above background,this paper proposes an expert recommendation algorithm based on knowledge mapping and collaborative filtering,which aims to alleviate the semi-automatic optimization of power consulting experts in the approval of power grid science and technology projects.Firstly,this paper constructs the power consulting expert knowledge map by bottom-up method.The extraction of triples in the atlas provides data support for experts’ recommendation.Then,this paper proposes a knowledge representation learning method based on ptranse model,which realizes the relationship link of knowledge map nodes through multi-step path on the basis of traditional trans e model.Experiments show that ptranse model can more accurately map entity relationship to low dimensional vector space.Finally,this paper uses a*algorithm to find the shortest path between the project to be approved and the experts on the knowledge map.and then calculates the expert fusion similarity matrix based on the idea of collaborative filtering to realize the intelligent optimization of power consulting experts.Combined with the expert recommendation algorithm based on knowledge map and collaborative filtering,the results show that ptranse model is better than mean rank and collaborative filtering Hit@10 The results show that the two indexes are better than the traditional translation model,which has better effect of entity vectorization;at the same time,the additive combination strategy of ptranse model is more suitable for the complex relationship network of the atlas described in this paper.In addition,in ptranse CF algorithm,the comprehensive ratio of the two expert similarity fusion is controlled at 0.7,and the nearest neighbor number k is controlled at 30,so the algorithm achieves the optimal efficiency.Under the optimal efficiency,the algorithm is obviously superior to the traditional collaborative filtering recommendation model and translation model,and the recommendation accuracy of the optimal efficiency reaches 75%.From the above results,we can see that ptranse CF model can adapt to the knowledge map network of power consulting experts,and can realize the semi-automatic selection of power consulting experts to a certain extent.