Abstract: Correctly identifying VN structure in Chinese can help to improve the accuracy of parsing.This paper proposes and validates the hypothesis that if the distributions of the contexts of VN combinations are similar，their structures are also similar.Combining the verb and the noun，the structural vector space model based on probability distributions is constructed to identify the VN structure.Experiments show that，without the use of other resources，the precision of this kind of method can achieve 95.2% and the recall can achieve 93.0%.

Key words: natural language processing, vector space model, VN structure, context

摘要： 正确识别汉语里的VN结构等基本名词短语可以帮助提高句法分析的准确率。提出并验证了如果动名组合的上下文词语的分布类似，那么它们的结构也类似的假设。结合动词、名词本身，构造了一种基于概率分布的结构向量空间模型，用于VN结构的识别。实验结果表明，虽然没有使用其他外部资源，该方法仍取得了理想的识别效果，精确率和召回率分别达到了95.2%和93.0%。

关键词: 自然语言处理, 向量空间模型, 定中（VN）结构, 上下文