摘要
句法分析是自然语言处理中的关键一环,目前的句法分析一般是依据句子中词的词性标记来进行的。而汉语单纯依据词性很难确定词之间正确的句法关系。在词类这个平面上进行句法分析存在着难以排除的结构歧义现象,因而使用语义知识排除结构歧义的方法更显重要。文中提出基于ATN(Augmented Translation Networks)的句法分析的新方法,对ATN进行相应的改进,利用《知网》的语义知识资源对其成分进行各种特性的标注。
Syntactic analysis is the key in nature language processing. It is generally carried on according to the morphological feature marks of a sentence, But only using the morphological feature of a word, it＇s very difficult to confirm the correct syntactic relations between the words in a Chinese sentence. There are structure ambiguous phenomenon when analyes a Chinese sentence syntactically. It is very difficult to get rid of the structure ambiguities, if simply using morphological features of the words. Therefore it is important to use semantic knowledge to get rid of the structure ambiguities. This paper puts forward a new method based on improved ATN （Augmented Translation Networks）. The method uses ＂Hownet＂ as the semantic knowledge resource to label the different composition in a Chinese sentence.
出处
《计算机技术与发展》
2006年第1期41-43,46,共4页
Computer Technology and Development
基金
陕西省教育厅专项科研基金(HD01302)
