Power User Service Recommendation Based On Hybrid Neural Model

With the popularity of information technology,the power industry with huge users has gradually upgraded,and the power service has begun to transform e-commerce model,but the current service still uses passive service methods,so introducing the recommended system into electricity services.The field is imperative,relying on the huge user group of the grid,the recommended system has the broad application prospects in the power industry.This paper studies in the recommendation of power service,first study for user demand prediction,to grasp user demand changes,and then combine the demand prediction model with the recommended model,propose a service of mixed neural network model Recommended models,by analyzing user needs,to improve service quality,improve service quality,improve enterprise competitiveness,this article works as follows:(1)In order to better serve the user,reduce the labor cost of the traditional service mode,consider the timing of user demand in the power service,that is,the user needs continuously changes over time,while considering the traditional cycle The problem exists in the neural network,thus combining the attention mechanism with the circulating neural network,proposes a circulating neural network model of combined attention mechanism,and is applied to the power service data set,the experimental results show that the model is in Precision The effects of four indicators such as Recall,F1-score,and Accuracy are significantly higher than the comparison model;(2)In order to comply with the shift trend of power services to e-commerce model,introduce existing e-commerce social information into electricity service,and improve in the previous work,introduce user social relationship attributes,and build power service map structures through social relationships.Data;and then considering that the traditional chart neural network computation cost is large,the introduction of the polymer sampling method is improved,and finally applied to the electrical service diagram structure data,the results show that the model is improved compared to the optimal comparison model,while To ensure credibility,the model is applied into the CORA public dataset,indicating that the model is improved compared to the optimal comparison model.(3)Recommended power services recommended by the user’s demand dynamic characteristics.In order to improve the demand for the introduction of social relations,the dynamic influence of user demand is considered,considering the needle neural network as polymerization operators,and constructs a mixed neural network recommendation model,which can be more comprehensive to user demand.Analysis,simultaneous add service information data on the original power service data,so that the model can obtain multi-data information,to recommend valid services for the user,and the experimental results show that the model is improved.