本文探讨了面向汉语信息处理的词语义项区分应该遵守的原则和方法。先界定了作为计算机自动词义消歧对象的多义词的范围;然后指出面对大规模真实文本,词语义项区分应具有可操作性,即应具有完备性和离散性;最后论证了上下文语境是计算机区分词语意义的最终凭借,信息处理用词语义项区分的依据应主要来自词语的句法行为。 This article explores the principles and methods that should be followed in the semantic classification of words for Chinese information processing. Firstly, the scope of polysemy as the object of disambiguation of computer automatic disambiguation is first defined. Then it is pointed out that in the face of large-scale real texts, the semantic categorization of words should be operable, that is, it should be complete and discrete. Finally, the context is computer The ultimate basis for distinguishing the meaning of words, the basis for distinguishing semantic items of information processing words, mainly comes from the syntactic behavior of words.