基于DBpedia的领域本体进化方法研究

Research on the Evolution Method of Domain Ontology Based on DBpedia

[目的/意义]领域本体是规范描述和语义化组织领域核心知识的模型,领域本体需要随着领域知识的变化而变化,自动化或半自动化的进化方法是领域本体动态更新的一个研究热点。在领域本体进化的相关研究中,大部分是基于非结构化的领域语料库,使用中文或英文分词工具来进行模式匹配实现本体进化,此种方法相对复杂,该研究提出利用结构化的数据作为基础实现本体进化。[方法/过程]DBpedia是从维基百科中提取的综合而庞大的结构化数据集,其数据结构能有效的和本体数据对接,为领域本体的半自动进化提供了一种可行的数据获取途径。采用DBpedia结构化的数据集作为领域本体进化源,提出了基于DBpedia的领域本体进化方法,主要步骤包括DBpedia信息抽取和优化、获取进化信息、本体的变更操作和一致性检查。[结果 /结论]以高速铁路动车组领域本体为实验对象,实现了动车组领域中英文本体的同时进化,该方法将为基于DBpedia的中文领域本体进化提供借鉴作用。

[ Purpose/Significance ] Domain ontology is a model that describes and organizes the core knowledge of the domain. Domain ontology needs to dynamically update with the changes of domain knowledge, so the automated or semi-automated evolution method is a hot topic. Most of the research in the ontology evolution is based on unstructured domain corpus, and the method using word segmentation tool to achieve ontology evolution is relatively complex. If we can use structured data as the basis to realize ontology evolution, it will great- ly improve the efficiency of ontology evolution. [ Method/Process] DBpedia contains billions of triples, which involves knowledge from diverse domains. The data structure of DBpedia is similar to the ontology＇s, and it provides a feasible way of data acquisition for the evolu- tion of domain ontology. Different from the method of ontology evolution that extracts new concepts from the unstructured text data, this paper proposes an evolution method of domain ontology based on the structured data of DBpedia. The main steps include DBpedia informa- tion extraction and optimization, obtaining the evolution information, ontology change operation and consistency checks. [ Result/Conclu- sion] This paper verifies that the method is feasible through the experiment on EMU domain of high-speed railway. This method could be used as reference to the evolution of Chinese ontology.