基于贝叶斯网络的本体映射研究-智能与分布计算实验室

基于贝叶斯网络的本体映射研究

姓名	刘涛
论文答辩日期	2007.01.26
论文提交日期	2007.02.02
论文级别	硕士
中文题名	基于贝叶斯网络的本体映射研究
英文题名	A study on ontology mapping based on bayesian network
导师1	李玉华
导师2
中文关键词	本体映射;不确定性;贝叶斯网络;迭代更新算法
英文关键词	ontology mapping;uncertainty;bayesian network;IPFP
中文文摘	于本体在表述语义方面的优势，越来越多的本体被开发出来，那么如何将本体集成就成为一个急需解决的问题，在集成过程中一个非常重要的步骤就是如何找到源本体和目标本体的映射关系。语义映射信息是在语义Web搜索中频繁使用的数据，良好的组织和描述方式有利于本体映射进程对信息的提取，从而改善本体映射的速度。如果描述方式选择好的话，可以让这些语义映射信息在语义Web上最大化共享。OWL由于在本体描述上的诸多优势成为最优选择。在本体映射中，概念的不确定性成为一个非常棘手的问题。在基于贝叶斯网络的不确定性建模框架中，扩展了本体标记语言，加入不确定性概念类来描述不确定性信息；源本体和目标本体被转换成贝叶斯网络，本体间概念的映射关系就可以通过先构造贝叶斯网络的条件概率表，然后用一种迭代更新算法找到映射关系。在本体概念结点数量较大时，经典的迭代更新算法IPFP在效率上存在不足，针对这个问题，改进算法I-IPFP中将约束按照结点间的关系分为本地约束和非本地约束两种类型，然后分别进行处理，有效提高了更新算法的效率。针对本体映射中概念的不确定性信息处理，设计了一个本体映射的原型，系统用Java语言开发，本体文件的构建使用开源工具Protégé，并利用了Jena2.2开源工具包作二次开发。
英文文摘	More and more ontologies are developed as the advantage of describing semantic, so ontology integration is becoming an imperative problem. A very important step in the process of ontology integration is to find the mapping information between the source ontology and target ontology. Mapping information is frequently queried in the searching process, So its organization, description forms and query methods have a great influence on the performance of semantic search engine. The Organization with Multi-value information system and the description form with RDF can simplify the information retrieval and share these mapping information on the Semantic Web widely. The mapping information query algorithm can exactly find out the correlative semantic mapping information by a set comparing method. How to deal with uncertainty is crucial in exact concept mapping between ontologies. This paper presents a new framework on modeling uncertainty in ontologies based on bayesian networks. In our approach, OWL is extended to add probabilistic markups for attaching probability information, the source and target ontologies are translated into bayesian networks, the mapping between the two ontologies can be digged out by constructing the conditional probability tables of the BN using a improved algorithm named I-IPFP based IPFP. The basic idea of this framework and algorithm are validated by positive results from computer experiments. The classic algorithm is insufficient when the ontology concepts’ number is big, the improved algorithm divide the constraints into local constraints and non-local constraints, and then hand them separately. This improves the algorithm effectively. Finally, the system is developed by java and the ontology is designed by the open-code tool named Protégé, ontology integration is applied by finding the mapping information between the ontology concepts.