需求概念图导引下的检索模型研究

发布时间：2018-01-25 01:07

本文关键词： 信息检索模型概念图需求分析相似度计算　出处：《上海交通大学》2013年硕士论文　论文类型：学位论文

【摘要】：信息检索模型的研究和发展已经历经70余载，在过去相当一段时间里，信息检索还只限于专业人员使用，模型的实现原理也比较简单，人们获取信息的途径并不依赖于信息检索，因此用户对于信息检索的需求还不是十分迫切。随着互联网的兴起，人们信息检索的需求也逐渐扩大。在纷乱庞杂的信息海洋中，如何准确地获取满足需求的信息也成为信息检索研究中一项刻不容缓的工作。当前大多搜索引擎在提供搜索服务的时候总有一些方面不如人意，究其原因，是因为它们把用户的需求简单地拆分成了若干个毫无关系的关键词，而没有把需求当作一个概念整体来看待，于是就丢失了关键词之间存在的语义信息。本文首先从概念图的相关理论研究入手，强调了概念分析在表征用户需求意图上的重要性。基于概念图的检索模型，通过需求概念图的标引来保持用户需求的概念内涵，在检索时融入概念图匹配和语义相似度计算的方法，从而提升检索的准确率。本文客观地分析了实现该检索模型的重重困难，同时重新考量了各项相关技术，创新性地提出了需求概念图导引下的检索模型。围绕这个模型，本文先着重讨论了需求概念图标引的方法，分析用户需求的相关特点，并结合词汇知识获取等方法，探讨E-A-V形式的概念图自动标引。其次，我们又详细介绍了相关语义相似度计算以及概念图相似度计算的方法，，对比了各自的优劣。最后是检索模型实现的各种细节，主要介绍了概念图实际应用时的实践经验，为今后概念图完全应用于信息检索，真正实现语义搜索提供一些有益的思路。
[Abstract]:The research and development of information retrieval model has gone through more than 70 years, in the past quite a period of time, information retrieval is only limited to the use of professionals, the implementation principle of the model is also relatively simple. People's access to information does not depend on information retrieval, so users' demand for information retrieval is not very urgent. With the rise of the Internet. People's demand for information retrieval is also gradually expanding. In the chaotic sea of information. How to accurately obtain the information to meet the needs has become an urgent task in the information retrieval research. At present, most search engines are always unsatisfactory in some aspects when providing search services, which is the reason. Because they simply split the user's requirements into several unrelated keywords and did not treat the requirements as a whole, they lost the semantic information that exists between keywords. This paper starts with the related theoretical research of concept map, and emphasizes the importance of concept analysis in representing the intention of user demand. The retrieval model based on concept graph is proposed in this paper. Through the indexing of the requirement concept map to keep the concept connotation of the user requirement, the method of concept map matching and semantic similarity calculation is incorporated in the retrieval. In order to improve the accuracy of retrieval, this paper objectively analyzes the difficulties of implementing the retrieval model, reconsiders the relevant technologies, and creatively puts forward the retrieval model guided by the requirement concept map. Around this model, this paper first discusses the method of conceptual icon citation of requirements, analyzes the relevant characteristics of user requirements, and combines the methods of acquisition of lexical knowledge. E-A-V form of concept map automatic indexing. Secondly, we also introduce the relevant semantic similarity calculation and concept map similarity calculation methods in detail. Finally, the details of the implementation of the retrieval model are discussed, and the practical experience in the practical application of the concept map is mainly introduced, so that the concept map can be fully applied to information retrieval in the future. The real implementation of semantic search provides some useful ideas.
【学位授予单位】：上海交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.3

【参考文献】