Chinese Learning of Semantical Selectional Preferences Based on LSC Model and Expectation Maximization Algorithm

摘要

Aiming at the situation of current Chineselanguage resources shortage ,this paper proposessemantically selectional preferences of unsupervisedlearning method, and presents a strategy of obtaining verbnounsemantic collocation in Chinese. An approach ofChinese semantic preference learning, which is based onLatent Semantic Clustering model and ExpectationMaximization Algorithm. First, the parameters areinitialized randomly. Second, a certain number of trainingiterations is performed until convergence. Each iterationconsists of expectation step and maximization step. Finally,the semantic association between verbs and nouns arecalculated as a measure of its matching probability. Thismethod can be used on Chinese without syntax-annotatedcorpora. Lots of experiment results show that LSC providesproper patterns of verb-noun collocation semantically. Thealgorithm converges quickly.