基于最大熵原理的汉语词义消歧

词义消歧是自然语言处理中亟待解决的一个关键问题，本文提出一种基于最大熵模型的有监督的机器学习方法，用于汉语词义消歧。该方法综合了词标记、词性、主题等上下文特征，并用一种统一的表示方法规范化特征形式，解决了多种不同特征之间的融合和特征的知识表示。实验对20个汉语高频多义词进行了测试，平均正确率为87％，验证了该方法的有效性。

Maximum Entropy-Based Chinese Word Sense Disambiguation

Word sense disambiguation is a crucial problem to be solved in NLP. A supervised machine learning method is proposed in this paper, which is applied in word disambiguation in Chinese. The method combines various features in context to disambiguate word senses. The features include annotations of words, parts of speech and sub- jects etc. And a uniform representation formalizes the features. In this way, the problem of synthesis among various features and knowledge representation of feature will be solved. 20 Chinese polysemous words are tested in our exper- iment. The result with average precision 87% shows that the method is effective.