语义关联关系的搜索和排序方法研究-智能与分布计算实验室

语义关联关系的搜索和排序方法研究

姓名	赵艳涛
论文答辩日期	2007.05.31
论文提交日期	2007.05.31
论文级别	硕士
中文题名	语义关联关系的搜索和排序方法研究
英文题名	A Study on Searching and Ranking of Semantic Association Relationship
导师1	卢正鼎
导师2
中文关键词	语义网;本体论技术;语义关联;语义关联搜索;贝叶斯网络;语义关联排序
英文关键词	Semantic Web;Ontology Technology;Semantic Association;Semantic Association Searching;Bayesian Network;Ranking Semantic Association
中文文摘	当今网络的一个主要目标就是信息共享，即不管什么样的平台、语言和协议，我们都能访问到需要的信息，这正是当代网络背后的商业需求。过去十年里网络搜索技术集中在显式存在的网络信息资源上，语义网的发展以及语义网体系中的层次架构使计算机处理隐式存在的语义信息，从而使信息资源检索时更有效率。对语义网上的语义关联关系搜索及其排序做了研究。语义关联的搜索即对本体中实体之间的语义关联进行搜索，一直以来，国内外对语义关联的概念没有权威的定义和说明。根据现有的资料，在前人研究的基础上，讨论了语义关联的含义。在某个知识领域中的两个实体，如果它们通过一个或多个属性直接连接在一起，或者是相似（相同或衍生）的属性间接连接在一起，就称之为语义关联。提出了一个初步的解决语义关联搜的方案，即利用贝叶斯网络。首先定义了语义关联搜索及其分类，接着定义了关联搜索的任务，然后将本体的分类作为贝叶斯网络的网络结构，利用实例的查询日志来估计网络的参数，当贝叶斯网络构建好后，文章给出了网络中关联搜索的解决方案。正如文档的排序对当今的搜索引擎至关重要一样，对语义网中搜索到的复杂关联关系进行排序也是以后的语义网搜索引擎的重要组成部分。研究设计了一种灵活的排序方法，它能够让用户获得更感兴趣和相关度更高的关联关系。另外，通过现实世界的数据对设计的排序方法做了一个评估测试来证明排序方法的有效性。
英文文摘	Information sharing is one of main target of current network, we can access the information we need, no matter what kind of platform, language and protocol, and this is the business requirement behind the network. So network searching focused on the standard of the information resource which guaranteed to be access in the past ten years. The development and hierarchy of semantic network can make the information searching more effective. In the thesis, the author makes some research on the searching and ranking of semantic association relationship. The searching of semantic association is to find out relationship which exists between two instances, there is no authoritative definition and explanation exist about semantic association. The thesis discusses the meaning of semantic association based on the existing information and research before. There are two entities in a knowledge field, if they connected by one or more properties, or connected by similar (homology or derived from) properties indirectly, then call it semantic association. The thesis proposes a preliminary solution for semantic association searching using Bayesian network. Firstly, the author defines the association search and its categorization, and then defines tasks in association search。In terms of Bayesian network, I take ontology taxonomy as network structure in Bayesian network. I use the query log of instances to estimate the network parameters. After the Bayesian network is constructed, we give the solution for association search in the network. Just as ranking of documents is a critical component of today’s search engines, the ranking of complex relationships will be an important component in tomorrow’s semantic web search engines. The thesis presents a flexible ranking approach which can be used to identify more interesting and relevant relationships in the semantic web. Additionally, I demonstrate my ranking scheme’s effectiveness through an empirical evaluation over a real-word data.