| The scale of business operations continues to expand,and there are more and more customers.The workload of customer service personnel has increased,but the service capacity is limited.Taking telephone customer service as an example,more than 60% of the problems in customer calls or online customer service representatives are common,and companies have to pay for repeated workloads.However,if robot service can be applied,the labor cost will be greatly reduced.At the same time,Internet content is exploding,making information search results mixed.In the field of health consultation,the number of patients has increased year by year,and China's medical and health conditions cannot keep up with the patient's medical needs.Therefore,how to use the information in the network to solve the current patient's health consultation problems through artificial intelligence and other new technologies in the computer field,The information construction in the medical field is an urgent problem to be solved today.In response to the above problems,this paper proposes a goal-oriented dialogue system that integrates retrieval and generation models and uses it in the field of health consultation.The system through the semantic analysis of the user's query on health consultation,from the large-scale information source reliable health consultation knowledge base learned the answer to the user to help them answer the health doubts.The goal-oriented dialogue system that combines retrieval and generation models mainly consists of three parts: problem retrieval module,semantic matching module and reply generation module.We have built a data set in the field of health counseling to train the models that will be used in these modules.At the same time,we demonstrate the effectiveness and superiority of the algorithms in these three modules through a series of experiments.By combining the retrieval model with the generation model,the overall performance of the dialogue system has been greatly improved. |
PDF Full Text Request