Design And Implementation Of Question Recognition And Text Similarity Module In A Customer Service Question Answering System

With the continuous development of human society and the continuous progress of science and technology,under the background of big data,people's technical needs for intelligent recognition of natural language become more and more urgent.Big data and artificial intelligence have penetrated into every aspect of daily life.The customer service industry also faces new challenges and changes due to the development of the Internet and the advent of the era of big data,which makes people have new expectations for the customer service industry.Natural language processing is an important branch of artificial intelligence to solve big data problems.QA(Question Answering)is an important application field of NLP(Natural Language Processing).In recent years,many fruitful research results have been obtained in the field of Natural Language Processing.This is a very practical direction,the future of search engines,is likely to be QA+reading comprehension,machines learn to read comprehension,understand each article,and then for the user's question,directly return the answer.Wisdom teeth technology co.,ltd.decided to launch an intelligent question-and-answer system for the customer service field,from the input of business field corpus to the output of the question-and-answer system automation system.To A certain extent,the construction of the question answering system relies on the recognition of text interpretation and user intention,and carries out QA pair extraction on corpus data.In practical application,it is inevitable that the machine cannot understand the meaning of the text and can't correctly identify Q and A,which results in the question answering system can't accurately extract all QA pairs.In addition,there are numerous ways to express a meaning in Chinese questions.As a general-purpose intelligent question-answering system,when the user's questions do not exist in the "question-answer"pairs,it is necessary to find the questions matching the user's questions in the existing collection of "question-answer" pairs and return the corresponding answers to the user.In order to solve the above two problems and realize the QA pair matching and recognition of the question answering system,it is very necessary to identify the question and the text similarity.Beijing wisdom teeth bochuang technology co.,ltd.decided to launch an intelligent question-answering system for the field of customer service,realizing the automatic construction from the input of business dialogue corpus to the output of question-answering database,and using an intelligent method to realize the intelligent return of the most relevant answers to customer service question-answering platform users' questions.This paper proposes two key technical steps to realize the question answering system based on the above ideas:question recognition and text similarity recognition,and builds an independent subsystem of question recognition and text similarity recognition.The purpose is as follows:first,the question and non-question can be correctly identified in the question answering system,and then the corresponding answer can be given.Second,for the new questions raised by the users of the question answering system,the most similar answers can be found in the existing questions and returned to the users as the answers of the question answering system,which also means that the same semantic questions in the question answering system should be answered with the same answers.Three is due to the question recognition and text similarity is one of the important steps,question answering system implementation greatly affects the quality and using effect of question answering system,and also can be used as a natural language processing tasks and exist independently,so set up questions identification and text similarity identification subsystem to question recognition and text similarity analysis and using the two independent task.This paper successfully completes the construction of the subsystem of question recognition and text similarity recognition.As a subsystem of the question answering system,the recognition of Chinese questions and the recognition of Chinese text similarity is an essential step for the realization of the question answering system.