基于推理信息的本体模块化方法

Reasoning Information Based Ontology Modularization

本体模块化在本体推理和复用等应用中有着极为重要的作用。怎么样将本体划分成小的模块是最基本的问题,目前本体模块化的工作主要集中在本体复用的目的上。在这篇文章中,我们提出了一种基于推理信息的本体模块化方法,该方法以提高推理的性能为目的。在基于同一最小推理集合内的公理之间内聚性将会增强的合理假设下,我们的模块化方法通过分析每次推理的过程,得到推理的最小推理集合,然后增强最小推理集合内公理1之间的内聚度,最后根据内聚度将本体划分成模块。在评估阶段,我们首先使用训练公理划分本体,然后通过测试公理来调查本体推理性能提高的程度。根据训练公理和测试公理所属范围的不同,我们使用了三组实验:训练公理和测试公理限定在同一较小范围,训练公理和测试公理不限定范围,训练公理和测试公理的范围实现三次不同的改变。实验的结果尤其是符合实际应用情况的实验三的结果证明了基于推理信息的本体模块化方法的有效性。

There is a growing need for applying the principle of modularity to representations of ontological knowledge in order to facilitate the scalability and reuse of ontology. How to partition ontology automatically is the primary issue, current research work has focused on re-use of ontology. In the paper, a reasoning information based ontology modularization method is presented for the purpose of enhancing the reasoning performance in ontology. According to the as- sumption that axioms which are in the same Minimal Reasoning Set have high cohesion,our method firstly calculates Minimal Reasoning Set of every reasoning task by analyzing the process of reasoning, then cohesion between axioms in the same Minimal Reasoning Set is increased; finally ontology is partitioned into modules according to cohesion. In the evaluation stage, training axioms are used to modularize ontology firstly, then some test axioms are carried to investigate the enhancement of reasoning performance. Three experiments are used to evaluate our ontology modularization approach, the results, especially the result of experiment three which is similar to practical situation prove the effectiveness of the reasoning information based ontology modularization method.