基于概念的信息搜索方法研究-计算机应用技术专业论文.docxVIP

华中科技 大学硕士 学 华 中 科 技 大 学 硕 士 学 位 论 文 I I 摘 要 随着 Internet 的快速发展和计算机的普及，人们已经进入了网络信息时代。但随 着信息高速增长，人们从网上查找和获取信息也就变得越来越困难，如何迅速、高 效地检索到最相关的信息是人们面临的一个重要问题。传统搜索引擎存在着基于关 键词机械匹配的局限，已不能满足用户的检索需求，因此基于语义的信息搜索已成 为信息搜索的一个研究热点。 针对目前基于语义的信息搜索存在同一概念但有不同表示形式的词汇而无法被 检索到的问题，一种基于概念的信息搜索模型被给出。该模型首先以知网为基础资 源，给出了一种概念扩展算法来从概念层次上扩展查询词，进而给出一种基于概念 的扩展词过滤算法，从语义角度滤掉低相似度的扩展词，最后给出一种基于概念扩 展的结果排序算法，来把概念扩展融入到结果排序算法中。 针对目前现有的词汇情感倾向分析方法是基于语义层次的，还缺少从概念层次 上去分析，一种基于概念的词汇情感倾向计算方法被给出。该方法以基于概念的相 似度为基础，给出了基于概念集的语义相似度、基于概念的情感倾向计算公式，并 进而给出其算法描述及其理论分析。 最后在以上方法研究的基础上，实现了原型系统并进行实验测试。结果表明概 念扩展查询方法提高了搜索的查全率，基于概念扩展的结果排序方法能够使查询结 果的查准率得到有效的保证，基于概念的词汇情感倾向计算方法有较高的准确率。 关键词: 语义搜索，概念扩展，结果排序，情感倾向 II II Abstract With the rapid development of Internet, people have entered an era of Internet information. At the same time, how to obtain the information efficiently that customers want from this enormous resources databases, is an important problem for all users. Traditional search engines based on keywords have the limitation of mechanical matching, and they cloud hardly meet the needs of customers now. Intelligent information retrieval system has become an important research area, in which retrieval system based on concept is one of the important methods. Aiming at solving the problem, which is that at present IR employing keyword-matching mode could not retrieve the other results with identical word meaning but dissimilar expression form, this thesis proposed a IR model based on concept. First in this paper, taking HowNet as a basic resource, this thesis proposed a kind of concept-expanding algorithm in order to extend the inquiry word on the concept level, and then we brought forward a filtering algorithm grounded on those words which have been extended at former step, for eliminating the extended words with low-relativity comparing to original word from semantic point of view. At last a result-sorting algorithm, which has been blended with the mechanism of concept-expanding, appeared. For the current word’s orientation analysi