【作者】 吴定峰；

【导师】 周国民；

【作者基本信息】 中国农业科学院 ， 作物信息科学， 2012， 博士

【摘要】 一般的信息检索系统存在检索效能低，检索结果答非所问等问题，其原因主要在于系统无法理解用户的语义，从而难以满足用户的检索需求。针对这一问题，人们提出了语义信息搜索的概念。最近，语义信息搜索的相关研究已逐渐成为信息检索领域的研究热点。然而现阶段的语义搜索研究面临着两大难题：一是没有找到一个有效的解析用户语义的手段，二是没有形成一个成熟的语义搜索模型。这两大难题已经严重阻碍了语义信息搜索研究的深入发展，成为这一领域急需解决的关键性问题。本文以情境理论为理论基础，以本体论为方法论基础，在信息检索过程中引入情境变量作为缩小语义开放性和准确理解用户需求的手段，利用本体将抽象的情境因素表达为信息检索系统可以读取和利用的具体变量，并最终发展成为一个以本体为基础的语义信息搜索模型。为了验证该模型的可行性，建立了一个果树语义信息搜索实验性系统，通过实验证明了该系统在检索效能和检索结果用户满意度方面具有较大的优越性。本文取得了如下三个方面的创造性研究成果：一是提出了基于情境变量的语义信息搜索框架。在这一框架中引入了用户的知识结构、用户承担的工作任务和信息环境这三类情境变量来精确识别用户信息需求，并提出了利用本体模拟用户知识结构和表达用户承担的工作任务的方法。二是在语义信息搜索框架基础上提出了一个以情境为导向以本题为基础的语义信息搜索模型，并重点论述了工作任务感知算法、工作任务表达算法和结果排序算法这三个核心算法，以及与这些算法相配合的本体知识库结构。三是结合果树领域本体，研制了一个果树语义信息搜索实验性系统。通过一个和百度搜索引擎的对比实验证明了该实验性系统在检索结果相关性、检索效能和智能程度上都有较大的提升，从而检验了语义信息搜索模型的可行性。更多还原

【Abstract】 General information retrieval system sxposed some obvious defects, such as low retrievaleffectiveness and Inappropriate answers. A very important aspect of the causes of this problem lies inthe information retrieval system can not understand the semantics of the user, which makes it difficult tomeet the user demands. To solve this problem, the concept of semantic information search has beenproposed.Since it proposed, the semantic information search is becoming a hot area of informationretrieval research. However, at this stage, the semantic search research is faced with two problems: First,did not find an effective means of parsing user semantic and locating the user demand. Second, has notformed a full-fledged semantic search model. These two problems has been a serious impediment to thein-depth development of the semantic information search, so, they are the key issues in this area need tobe resolved.Based on the situational theory, this study combinated the contextual factors as an effective meansof narrowing the user semantic openness and positioning the means of user needs. The ontology hasbeen used to exchange the abstract contextural factors into some specific various which can be read anduse by information retrival systems. Based on these various, a semantic search model has been built. Totest the model, an experimental system was developed which has been proved has a greaterimprovement by an experiment.Three contributions have been made:The first is proposed a context-orinted semantic search frame. Three context variables affecting theinformation retrieval process has been selected into the framework from many context factors. They areuser knowledge structure, task and information environment. Further more, the method of expressingthe context variables was proposed.The second is made a technical program to use the semantic search framework in semanticinformation search. Based on the semantic search framework, a context-orinted semantic search modelwas proposed. In this technical solution, the task-aware algorithm, the tasks expression algorithm andthe results sorting algorithm are discussed and the ontology structure compatible with these algorithmsis expounded.The third is developed a pomology semantic search system. This system has been proved has agreater improvement in the search results appropriate level, retrieval performance and the degree ofintelligence by a comparison test with Baidu search engine. So, the feasibility of the semantic searchmodel has been tested.更多还原