摘要
本文从术语研究的语言学视角,提出将形态特征同现有术语抽取方法相融合的多词术语自动抽取策略,并通过抽取实验对该策略进行了评估。结果表明,形态特征和基于句法规则方法相融合能够显著提高术语的自动抽取效率。研究同时发现,形态特征值能够有效地区分术语和非术语。本研究不仅揭示了语言学知识在术语自动抽取中的作用,同时为以语言学为支撑的自然语言处理研究范式提供了有力支持。
From the linguistic dimension of terminology, the present paper proposes a new strategy for automatic multi-word term extraction by incorporating the morphological features into the current approaches. The experiments indicate that the incorporation of morphological features into the rule-based approach is robust in improving the efficiency of automatic term extraction. The score of morphological features is effective in distinguishing terms from non-terms. The study not only reveals the significance of linguistic knowledge in automatic term extraction but also provides strong support for the linguistics-underlain paradigm of natural language engineering.
出处
《外语电化教学》
CSSCI
北大核心
2013年第2期35-40,共6页
Technology Enhanced Foreign Language Education
基金
国家社科基金项目"基于平行语料库的术语自动抽取及双语术语词典编撰研究"(11BYY053)的阶段性成果之一
国家留学基金委公派访问学者(含博士后研究)项目资助
关键词
术语自动抽取
形态特征
多词单位
融入策略
Automatic Term Extraction
Morphological Features
Multi-Word Units
Incorporation Strategy