Eliminating misunderstanding of one word with multi-semantic based on context
Lin Hai-wen · Jisuanji gongcheng yu sheji · 2006
To eliminate the misunderstanding of one word with multi-semantic, the traditional methods, such as the method based on the statistical relevance of the words in the same sentence and that based on the instances, have pretty effects, but some shortcoming exist also. Thereby, the method made use of the context is proposed, as follow: Firstly , adding the parameter of the distance between the words in the same sentence , which influences the statistical relevanceof the words, at the same time, taking into accountthe frequently- selected semantic, affects the current semantic of a word; secondly, to solve the sparse data problem in the method based on instances, a method that combines the select-restriction based on ontology and the calculation of the probability is proposed. The experiment shows that the methods are effective.