基于领域本体中文自动问答系统相关技术的研究与实现

发布时间：2018-04-03 04:25

本文选题：问答系统　切入点：本体　出处：《华东理工大学》2013年硕士论文

【摘要】：随着互联网技术的发展,信息量暴增,给人们的生活发生了翻天覆地的变化。现在,人们已经习惯于在互联网上获取各种各样的信息。这主要归功与搜索引擎技术的发展。然而,传统的搜索引擎仍然有一些缺陷。比如,用户只能通过关键字词进行检索,这并不能充分表达用户的搜索意图；又比如,传统索索引擎返回许多相关的候选结果,待用户从中找到其目标结果,这样的召回率往往很低,用户体验较差。针对以上问题,自动问答系统运用而生。用户使用自然语言问句向自动问答系统提问,系统返回的是对问句最直接最简单的答案。本文首先对现在已有的问答系统中的技术理论进行了分析,阐述了各个模块所使用技术的优势和不足。然后,参照国外一些本体构建工程,按照这些本体工程提出的构建方法论和经验,构建了小型的零售领域本体知识库,用于检索面向受限领域的知识。以本体在问答系统中的应用为出发点,提出了基于零售领域本体库的问答系统的答案抽取方法。用户使用自然语言问句向系统提问,经过分词、去停用词、语义标注等步骤,使用浅层语义分析技术对问句进行分析,得到问句中的已知和未知信息,在此基础上生成问句向量。最后使用SPARQL查询语言从本体库中查找问题答案。由于是直接查找问题的答案,有效地提高了系统的召回率,改善了用户体验。基于以上理论,设计并实现了面向零售领域的自动问答系统模型。通过应用验证了本文提出的相关技术,证明了本系统相关理论的可行性。
[Abstract]:With the development of Internet technology, the amount of information increases dramatically.Nowadays, people are used to getting all kinds of information on the Internet.This is mainly due to the development of search engine technology.However, the traditional search engine still has some defects.For example, users can only search by keywords, which does not fully express the user's search intention. For example, the traditional Sosso engine returns many related candidate results, and the user finds the target result from the search engine.Such recall rates tend to be low and the user experience is poor.In view of the above questions, the automatic question answering system is used.Users use natural language questions to question the automatic question answering system, which returns the most direct and simple answer to the question sentence.In this paper, the technical theory of the question and answer system is analyzed, and the advantages and disadvantages of the technologies used in each module are expounded.Then, referring to some overseas ontology construction projects, according to the construction methodology and experience proposed by these ontology projects, a small retail domain ontology knowledge base is constructed, which is used to retrieve the restricted domain knowledge.Based on the application of ontology in Q & A system, an answer extraction method based on retail domain ontology library is proposed.Users use natural language questions to ask questions to the system, through word segmentation, deactivation words, semantic tagging and other steps, using shallow semantic analysis technology to analyze the question sentence, and get the known and unknown information in the question sentence.On this basis, the question vector is generated.Finally, the SPARQL query language is used to find the answers from the ontology library.It can improve the recall rate of the system and improve the user experience.Based on the above theory, an automatic question answering system model for retail field is designed and implemented.The feasibility of the related theory of the system is proved by the application of the related technology proposed in this paper.
【学位授予单位】：华东理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.3

【引证文献】