电子商务领域本体概念及概念间关系的自动抽取研究

Research on Automatic Extraction of Dontology Concept and Its Relation in E-commerce

【目的/意义】使用人工或常用软件工具获取本体概念及概念间关系已无法满足自媒体环境下大数据的本体构建及维护的要求,本文尝试用自动或半自动方式予以实现。【方法/过程】对电子商务领域原始语料进行分句、分词等预处理,构建领域语料库;使用基于语言学以及统计学的方法提取电子商务领域本体概念,同时提出基于混合策略的本体概念抽取方法;然后使用基于语言学、聚类的方法提取电商领域本体概念与概念之间存在的分类关系,使用基于关联规则挖掘的方法抽取其本体概念与概念之间存在的非分类关系。【结果/结论】将文本挖掘与本体构建结合起来,提出了领域本体概念及概念间关系自动抽取与本体构建方法,实验表明使用本体框架Jena利用此方法可以自动构建电子商务领域本体,并可将其应用到通用的语义检索系统中。

[Purpose/significance] The concept of ontology and the relationship between concepts can＇t meet the require- ments of ontology construction and maintenance of big data in the We Media environment. This paper tries to implement it automatically or semi automatically. [Method/process] Primitive data in the field of electronic commerce preproeessing of clause and participle are carried out to construct domain corpus; then use linguistics and statistical methods based on the extraction of e-commerce domain ontology concept, and puts forward the concept of ontology extraction method based on hy- brid strategy; the taxonomy relation between its ontology concepts is extracted based on linguistics and clustering methods, the non- taxonomy relation is also extracted by the means of association rule mining. [ Result/conclusion ] This paper, based on the combination of text mining and ontology construction, puts forward the concept and the relationship between con- cepts of domain ontology automatic extraction and ontology construction method, and experiments show that the use of ontol- ogy framework Jena can automatically construct the electronic commerce domain ontology, and can be applied to the gener- al semantic retrieval system.