摘要
国际上,隐喻在思维及语言中所处的中心地位正逐渐引起人工智能研究者的重视。但在国内学术界,还鲜有开展隐喻计算化这方面研究的;实际上,作为异常用法的隐喻现象是自然语言中的普遍情况,因此隐喻问题若得不到很好的解决,将成为制约自然语言理解和机器翻译的瓶颈问题。本文结合相关的隐喻理论基础,根据不同的计算路线对已有隐喻理解计算模型进行分类,包括基于语义优先方法、基于知识表示的方法、基于逻辑的方法和基于统计语料库的方法,并在分析这些方法的适用范围和优缺点的基础上,对隐喻的计算理解方法以及面向汉语的隐喻理解计算模型研究提出了展望和建议。
Metaphor is prevalent in natural language. Researchers have realized that it is the focus of mind and language mechanism. The comprehension of metaphor by machine will be a bottle-neck problem in natural language understanding and machine translation. In this paper, current available computational models for metaphor comprehension are reviewed. According to their computational methods, the models are divided into preference semantics based, metaphorical knowledge based, logic based, and corpus based approaches. The advantage and limitation of each model are also analyzed. At last, prospect and advice to computational model for metaphor comprehension are proposed. As a conclusion, the research of computational model for Chinese metaphors is important for Chinese information processing.
出处
《计算机科学》
CSCD
北大核心
2006年第8期178-183,共6页
Computer Science
基金
国家自然科学基金资助项目(60373080)
关键词
隐喻
自然语言理解
优先语义学
计算模型
Metaphor, Natural language understanding, Preference semantics, Computational model
