摘要
针对自然语言处理领域词义消歧这一难点,提出一种新的汉语词义消歧方法。该方法以《知网》为语义资源,充分利用词语之间的优先组合关系。根据优先组合库得到句中各个实词与歧义词之间的优先组合关系;将各实词按照优先组合关系大小进行排列;计算各实词概念与歧义词概念之间的相似度,以判断歧义词词义。实验结果表明该方法对于高频多义词消歧是有效的,可作为进一步结构消歧的基础。
WSD（word sense disarnbiguation）is a difficult field in natural language processing. This paper puts forward a new Chinese WSD method. This method imposes HowNet as the semantic knowledge resource, meanwhile fully uses PER between words. Firstly,get PCR between the words through PCR database. Secondly, according to the size of the PCR, these notional words are put in order. Lastly, calculate the similarity degree between each concept of notional words and the concept of a word of different meaning, to obtain the right meaning of the word of different meaning. The experiment result proves this method to be effective for high frequency WSD. It can be the foundation of further structure disambiguation.
出处
《计算机技术与发展》
2006年第3期22-25,共4页
Computer Technology and Development
基金
陕西省教育厅专项科研基金资助项目(HD01302)
关键词
词义消歧
优先组合关系
词关联
相似度
知网
WSD
prefexence combination relation
word association
similarity
HowNet
