【摘要】 词义消歧是自然语言信息处理领域的基础研究,对自然语言信息处理领域的研究至关重要。为解决词义消歧中提取关联词语不精确进而影响词义消歧正确率的问题,该文依存句法模板设计了5种复合特征模板,并结合最大熵模型进行训练。实验证明,使用该复合模板,不仅降低了计算复杂度,而且提高了词义消歧的性能。对500余条复句进行词义消歧,取得了较好的词义消歧正确率。更多还原

【Abstract】 Word sense disambiguation is a basic research in the field of natural language information processing. It is very important for the study of natural language information processing. In order to solve the problem of inaccuracy of word association in word sense disambiguation,this paper proposes a dependency syntax template,and designs five kinds of compound templates,which are combined with the maximum entropy model. Experiments show that the proposed method can reduce the computationalcomplexity and improve the performance of word sense disambiguation. More than 500 complex sentences are disambiguation and the correct rate of word sense disambiguation is higher.更多还原