基于知网的汉语普通未登录词语义分析模型

提出一种基于知网的汉语普通未登录词语义分析模型,该模型以概念图为知识表示方法,以2005版知网为语义知识资源,首先参照知网知识词典对普通未登录词进行分词;然后综合利用知网中的知识词典等知识,通过词性序列匹配消歧法、概念图相容性判定消歧法、概念图相容度计算消歧法及语义相似度计算消歧法对中文信息结构进行消歧;最后根据所选择的中文信息结构生成未登录词的概念图,从而实现未登录词的语义分析。该模型在语义分析过程中一方面确定了未登录词中每个已登录词的词义,另一方面构造了该未登录词的语义信息,实验结果证明它可以作为普通未登录词语义分析的原型系统。

HOWNET-BASED SEMANTIC ANALYSIS MODEL FOR GENERAL UNKNOWN CHINESE WORDS

A HowNet-based semantic analysis model for general unknown Chinese words is put forward.It uses conceptual graphs as the method of knowledge representation and imposes HowNet version 2005 as the knowledge resource of the semantics.First,the model segments the general unknown Chinese words into some known words referring to the knowledge dictionary in HowNet;then by synthetically using the knowledge in HowNet such as the knowledge dictionary and so on,it disambiguates the Chinese message structures through the disambiguation methods of part of speech lists matching,conceptual graphs compatibility judgment,conceptual graphs compatibility computation and semantic similarity computation;Finally,it creates the conceptual graph of unknown Chinese words according to the selected Chinese message structure,and consequently completes the unknown Chinese words semantic analysis.In the model,the meaning of every known word in unknown words will be confirmed during semantic analysis on the one hand,and on the another hand the semantic information of the unknown words will be constructed.Experimental result proves that the model can be a prototype system for general unknown Chinese words semantic analysis.