基于规则的学术概念属性抽取  

Extraction of Academic Concept Attribute Based on Rules

作　　者：丁君军[1] 郑彦宁[1] 化柏林[1] 

机构地区：[1]中国科学技术信息研究所,北京100038

出　　处：《情报理论与实践》2011年第12期10-14,33,共6页Information Studies:Theory & Application

基　　金：国家自然科学基金项目"基于句子匹配分析的知识抽取研究与实现"的研究成果之一;项目编号:70803048

摘　　要：学术文献作为科研成果的结晶,蕴含着丰富的知识,属性抽取作为知识抽取的一种应用,是把不同学术文献中对于某一学术概念的属性描述集中起来,用来构建属性描述知识库。本文通过人工构建规则的方法,形成了属性抽取的九大类描述规则,并对属性描述进行了数量关系和情感信息方面的分析,然后对学术概念属性抽取系统进行设计和实现,并对《情报学报》2007年和2008年的文章进行抽取实验,对系统抽取的结果和人工标记的结果进行了测评,并进行了实验结果的分析。As the crystallization of achievements in scientific research, academic literatures contain a wealth of knowledge. Attribute extraction is an application of knowledge extraction, which collects the attribute description of a certain academic concept in different academic literatures to construct the attribute description knowledge base. By the use of the method of formulating rules manually, this paper forms nine major categories of rules for attribute ex- traction. The paper analyzes the quantity relationship and emotional information of attribute description. Then it designs and constructs the academic concept attribute extraction system. The paper performs an extraction experiment on the articles published in the ＂Journal of The China Society for Scientific and Technical Information＂ from 2007 to 2008, and evaluates the results extracted by the system and the rcsuhs marked manually. Finally, the paper makes an analysis of the experimental results.

关 键 词：知识抽取 规则构建 学术定义 概念 

分 类 号：G350[文化科学—情报学]