Research And Implementation Of User Demand Modeling Method Oriented To Intelligent Consulting Service

As a new product of the information age,intelligent consulting service is developed from traditional manual consulting service and based on data mining,machine learning and other technologies.For users,traditional consulting service is a social activity with a mentality of seeking knowledge,learning process,and question and answer forms;for service personnel,it is a task of understanding users,judging user demands,and answering questions.In actual consulting services,there are not all users having sufficient professionalism to support to ask specific and detailed professional questions.Therefore,in the fuzzy consulting scenario,analyzing user demands,returning the consulting results suitable for the user,and realizing the personalization of consulting services is a very important task for the realization of intelligent consulting services.Since intelligent consulting service are oriented to vertical fields and are still in the initial stage,there are few actual application data and it is difficult to effectively analyze user demands.In view of the above problems,this paper focuses on the technical consulting services of high-speed train motor operation and maintenance,and the fuzzy consulting that users fail to ask specific questions,and mainly conducts the following research:(1)Based on the analysis of factors affecting user demands in a specific field and the research results in the fields of psychology and pedagogy,the user demand model with a three-layer structure is proposed.Firstly,the model realizes user personalization by constructing user profile which is analyzing user learning ability and consulting interest.Then,based on the human learning law,that is,learning accumulation and memory forgetting law,the user demand analysis strategy and update method are proposed to analyzing and fitting the user demands;Finally,a three-dimensional representation model of user demands is constructed by gridding the knowledge domain,and the abstract concept of "user demand" is materialized.(2)Based on the analysis and construction of the user demand model for intelligent consulting service,a set of model evaluation criteria and optimization scheme is proposed.Based on the user feedbacks and the recall rate after promotion,the evaluation standard checks the actual application of the model,and adjusts the user demand analysis strategy through the optimization scheme to improve the fitting effect of the entire model to user demands.(3)Based on the survey results and the simulation user data set reasonable design rules,constructing user profile includes constructing simple rule-based labels,the comparison experiments between multiple classification models based on Light GBM and Random Forest for constructing learning ability label,the comparison experiments between the consulting interest-specific model and the general model based on binary classification models such as SVM and BP-MLP for constructing consulting interest label.All of the experiment achieved good results and verified that the consulting interestspecific models were better than the general model.Based on the comprehensive design and simulation of user consultation scenarios and feedbacks,the effects of user demand analysis strategy and updating method were analyzed and displayed.In addition,the effectiveness and feasibility of future practical applications of the model were discussed in combination model evaluation criteria and optimization scheme.