面向引用关系的引文内容标注框架研究

A Deep Scientific Literature M ining-Oriented Framew ork for Citation Content Annotation

引文内容分析能够帮助揭示文献引用关系的深层语义内涵。本文梳理了目前已有的引文内容标注体系,归纳出构建引文分类体系的三个主要维度,即引文功能,引文重要性,情感倾向。以支持文献引用关系分析为目标,针对引文内容分析设计出一个引文内容标注框架,其中包括揭示引文关系抽象性质的引文分类标注体系,描述被引文献具体内容的引用对象标注体系,以及记录引文客观特征的引文属性标注体系。具体的标注实验体现了该标注框架的可用性。

Citation content analysis can reveal the connotative meaning of citations. This article organizes existing schemes of citation content analysis. Three main dimensions of designing citation classification scheme have been concluded. In order to support automatic scientific literature mining, this paper proposes a framework for citation content annotation, which contains a citation classification annotation scheme indicating literature＇s abstract attributes, a cited object annotation scheme describing literature＇s specific content, and a citation attribute annotation scheme recording citation＇s objective features. A pilot annotation shows the usability of proposed annotation framework. 1 fig. 6 tabs. 56 refs.