中文学科术语相关语义关系获取方法研究

发布时间：2019-03-29 08:25

【摘要】：文章对中文非结构化文本中(半)自动获取学科术语的相关语义关系进行了探讨,以寻找行之有效的获取方法。从CNKI获取"数字图书馆"学科领域文献,通过术语抽取、动词抽取、向量空间模型构建、双重关联规则分析和规则评价获得了具有较强关联的术语对以及作为关联标签的动词,从而获取了学科术语的相关语义关系。该获取方法与其他方法相比,具有较高的可行性和有效性,并对术语的相关语义关系进行了有效性和实用性的评价,提高了获取的准确率。但文章也存在一定的局限性,在对术语相关语义关系的有效性和实用性进行评价时,指标的选择和阈值的确定存在人工干预,具有一定的主观性。
[Abstract]:This paper discusses the semantic relationship of automatic acquisition of disciplinary terms in Chinese unstructured texts in order to find an effective way to obtain them. Through terms extraction, verb extraction, vector space model construction, double association rule analysis and rule evaluation, we get the term pairs with strong relevance and verbs as the associated label through the acquisition of documents in the field of "Digital Library" from the CNKI, and through the terms extraction, verb extraction, vector space model construction, double association rule analysis and rule evaluation. Thus, the related semantic relations of the subject terms are obtained. Compared with other methods, this method is more feasible and effective, and evaluates the related semantic relations of terms effectively and practically, so as to improve the accuracy of the acquisition. However, there are some limitations in this paper. When evaluating the validity and practicability of the semantic relations related to terms, manual intervention exists in the selection of indicators and the determination of threshold values, which has a certain degree of subjectivity.
【作者单位】：南京大学信息管理学院;江苏省数据工程与知识服务重点实验室;
【基金】：江苏省社会科学基金一般项目“领域术语语义关系自动获取研究”(项目编号:15TQB009) 国家自然科学基金青年项目“面向学术资源的TSD与TDC测度及分析研究”(项目编号:71503121)研究成果之一
【分类号】：TP391.1

【参考文献】