摘要
文章提出一种改进的关联规则方法,用于抽取文本中的非分类关系。首先利用基于上下文的术语相似度获取方法得到术语间的相似度权重,再通过加入谓语动词的关联规则算法计算,结合搜索引擎技术得到候选关系对集合,并通过置信度和支持度的对比分析,抽取最终的非分类关系结果,最后对测试数据进行实验,并对结果进行分析。
An improved method of association rules used to extract the non-taxonomic relationships from the text is proposed. Firstly, the weight of the similarity between the terms is extracted by computing the similarly between the terms based on the context. Secondly, the collection of candidate relationship pairs is obtained by computing the association rules with verbs and by the use of search engine technologies, and the final non-taxonomic relationship results are extracted by comparative analysis of confidence and support. Finally, an experiment is carried out on test data, and the results are analyzed.
出处
《情报理论与实践》
CSSCI
北大核心
2011年第12期121-125,共5页
Information Studies:Theory & Application
基金
国家社会科学基金项目"面向语义网本体的知识管理研究"的成果
项目编号:09CTQ010
关键词
关联规则
非分类关系
本体
association rules
non-taxonomic relationship
ontology