Research And Implementation Of Court Customer Service Question And Answer System Based On Intention Recognition

With the drawbacks of traditional search engines constantly revealed,the development of the question and answer system has received more and more attention and rapid development.At present,as a kind of question and answer system,the intelligent customer service system is also developing rapidly and widely used.In the fields of e-commerce,finance,medical care and home appliances,intelligent customer service robots basically replace manual customer service,and serve users in real time and efficiently.Living in a legal society,the public and the courts have close ties,and court customer service plays a very important role.Manual customer service has been unable to meet the needs of the public,and it is urgent to replace it with smart customer service.In the court intelligent customer service system,the accuracy of question matching directly affects the accuracy of the system.High-precision question matching requires a deep learning technique to learn a large number of corpora.When the corpus is insufficient,it is difficult to achieve high-precision question matching.In order to ensure the accuracy of the system,problem understanding technology must become a strong pillar.It is the core idea of this paper to identify the user intent contained in the question and to search and match the most similar questions according to the intention.User intent recognition consists of two parts: one is to identify the type of user question,and the other is to identify the specific intent in the question.When the range of question sets is relatively concentrated,there is a high degree of word repetition between the questions,which greatly increases the difficulty of calculating the similarity of the questions.In order to solve this problem,this paper proposes to identify the question type first,and then only calculate the similarity between the same type of questions,so as to reduce the difficulty of measuring the similarity of the question.In this paper,a question-type classification model based on multi-feature fusion mining is proposed.Using the characteristics of word vector and part of speech,the two-way long-term memory network with attention mechanism is used for classification learning.Using the data of the court customer service compiled by hand,the experimental results show that the classification algorithm has a better improvement than other algorithms,and also significantly improves the accuracy of the question and answer system.For questions with different intentions,it is meaningless to calculate the similarity,and it is directly judged as not knowing.Therefore,identifying the specific intent of the user question will further reduce the difficulty of calculating the similarity of the question.In this paper,multi-label classification method is used to identify multiple intents in questions,and a multi-label classification model based on co-occurrence matrix is proposed.The relationship between categories is fully utilized.Experimental results show that the algorithm is improved compared with other algorithms.Improve the accuracy of the Q&A system.Finally,the paper completed the design and implementation of the court customer service question and answer system based on intention recognition.Build a complete customer service Q&A system that can accurately answer user questions.