基于UMLS专家词典与工具的词形归并算法研究  

Investigation of Algorithm for Lemmatisation Based onUMLS SPECIALIST Lexicon and Lexical Tools

作　　者：李晓瑛[1] 李丹亚[1] 胡铁军[1] 

机构地区：[1]中国医学科学院医学信息研究所,北京100020

出　　处：《情报科学》2013年第4期134-138,共5页Information Science

基　　金：国家科技支撑计划(2011BAH10B02)

摘　　要：在简述词形归并(原形化处理)基本目标的基础上,着重分析UMLS专家词典的构建方式与词典工具的核心功能,以及Norm原形化工具的处理机制;在此基础上,提出一种词形归并算法及Norm处理缺陷的修正办法,并收集医学词表数据进行算法测试与验证;此外,本文算法与经典的Porter算法进行了多方面的深入比较研究。Based on the brief description of the basic objective of lemmatization （normalization）, the mech- anism of developing UMLS SPECALIST lexicon and the key performance of the lexical tool, especially the working procedure of Norm, were carefully studied. Then an algorithm for lemmatization and the strategy to correct the defect of Norm processing were proposed, followed by testing and verification using the lexi- cal data from medical vocabularies. Besides, the well-known Porter method was used for algorithm com- parison.

关 键 词：词形归并 UMLS专家词典 Norm原形化工具 

分 类 号：G254[文化科学—图书馆学]