基于本体的自动问答客服系统研究

发布时间：2018-03-01 02:09

本文关键词： 本体问答系统语义网推理语义相似度知网　出处：《青岛理工大学》2012年硕士论文　论文类型：学位论文

【摘要】：随着社会经济的发展、企业改革的深入和人们生活水平的提高，企业正面临着良好的发展机遇，但同时企业和客户对服务提供与消费提出了新的要求：企业需要随着市场环境的变化调整自己的战略，及时、以最短时间、以最好的服务满足顾客的需求；同时客户要求更友好、方便、快捷的使用企业提供的服务。为此，企业需要通过不断改进客户服务体系来提高客户服务质量。然而在互联网发展的时代，传统的电话呼叫客服系统需要大量的人力和物力，企业为此需要大量的投入；现有的在线客服系统不能很好的理解用户意图，同时也不具备完善的知识体系，向用户提供所需信息。基于以上需求，本文从用户端和问题端两个不同的角度解决用户需求，针对手机领域对手机构建本体，结合自然语言处理、信息检索和语义网技术，一方面采用自然语言处理对用户请求问句进行处理，然后进行语义相似度计算和本体推理查询，从语义角度理解用户意图；另一方面结合本体，将手机领域知识构建到本体中，通过本体的良好结构实现知识的共享和组织；最后结合现实需求，设计与实现问答客服系统，以友好的自动问答方式与用户进行交互。针对基于本体的自动问答客服系统，本文主要工作和研究成果如下：（1）基于本体的自动问答客服系统的系统框架设计与实现。本文结合需求和当前中文问答系统的不足，设计与实现基于本体的自动问答客服系统框架，系统具有友好的UI用户交互，知识库和用户数据管理以web服务接口的方式提供。（2）手机领域知识本体的构建。本文结合手机领域知识的特点，以Protégé构建了手机本体，，将手机相关的软硬件、故障、现象、解决方法等组织到本体结构上，为后续的推理，实现知识的共享。（3）用户问句语义理解。本文问答客服系统区别于传统的搜索引擎，用户通过交互页面以自然语言形式输入问句。系统对用户问句结合分词、去停用词、同义词扩展、本体的语义扩展等自然语言处理，然后基于知网和构建的手机本体进行语义相似度计算，从语义角度理解用户问句。（4）基于本体的查询与推理。本文问答客服系统通过Jena解析构建的手机本体文件，用户问句经自然语言处理后，结合本体和定义的推理规则文件进行推理查询，引导用户进行问答，更好地获取用户问句相关的FAQ问题和答案。通过以上相关工作的研究，本文设计与实现了基于本体的自动问答客服系统。结合本文对问答系统的测评，相较于传统的关键词检索方法和常见的基于知网的问答系统，本文采用的基于本体语义相似度计算+本体推理查询方式，提升了查询结果的准确性，验证了系统的有效性，很大程度上能够理解用户的意图，本体的推理也能很好的引导用户进行问答，完全满足当前实际应用系统的需要。
[Abstract]:With the development of social economy, the deepening of enterprise reform and the improvement of people's living standard, enterprises are facing good opportunities for development. But at the same time, enterprises and customers put forward new requirements for service provision and consumption: enterprises need to adjust their strategies with the changes of market environment, timely, in the shortest time, with the best service to meet the needs of customers; At the same time, customers require more friendly, convenient and quick use of the services provided by enterprises. Therefore, enterprises need to improve customer service quality through continuous improvement of customer service system. However, in the era of Internet development, The traditional telephone call customer service system needs a lot of manpower and material resources, and enterprises need a lot of investment for this; the existing online customer service system can not understand the user's intention very well, and at the same time, it does not have a perfect knowledge system. Provide the user with the required information. Based on the above requirements, this paper solves the user's needs from two different angles of client and problem, constructs the ontology of mobile phone in the field of mobile phone, combines natural language processing, information retrieval and semantic web technology. On the one hand, natural language processing is used to process user request questions, then semantic similarity calculation and ontology reasoning query are carried out to understand user intention from semantic point of view; on the other hand, mobile phone domain knowledge is constructed into ontology by combining ontology. Through the good structure of ontology, knowledge sharing and organization are realized. Finally, a Q & A customer service system is designed and implemented to interact with users in a friendly automatic Q & A mode. The main work and research results of this paper are as follows:. This paper designs and implements the framework of automatic Q & A customer service system based on ontology, which is based on ontology, combined with the requirement and the deficiency of Chinese Q & A system. The system has friendly UI user interaction, knowledge base and user data management are provided by web service interface. 2) the construction of mobile phone domain knowledge ontology. This paper combines the characteristics of mobile phone domain knowledge, constructs mobile phone ontology with Prot 茅 g 茅, organizes mobile phone related hardware and software, faults, phenomena, solutions and so on to ontology structure, for the subsequent reasoning. Knowledge sharing is realized. This system is different from the traditional search engine. Users input questions in the form of natural language through interactive pages. The system combines questions with participle to stop words and expand synonyms. Ontology semantic extension and other natural language processing, and then based on the knowledge network and mobile phone ontology semantic similarity calculation, from the perspective of semantic understanding of user questions. 4) Ontology-based query and reasoning. In this paper, Jena is used to analyze the mobile phone ontology files. After the user questions are processed by natural language, the user questions are inferred and queried by combining ontology with the defined reasoning rules file. Guide users to ask questions and get better FAQ questions and answers. Through the research above, this paper designs and implements the automatic Q & A system based on ontology, and compares with the traditional keyword retrieval method and the common question-answering system based on knowledge net, combined with the evaluation of the question and answer system in this paper. The ontology reasoning query method based on ontology semantic similarity calculation is adopted in this paper, which improves the accuracy of query results, validates the effectiveness of the system, and to a large extent can understand the intention of users. Ontology reasoning can also guide users to ask questions and answer questions, which can meet the needs of current practical application system.
【学位授予单位】：青岛理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP391.1

【引证文献】