Abstract: This paper uses neighborhood rough set theory to the attributes reduction of oil-bearing reservoir and puts the residuary attributes reduction as the input variables of support vector machine to carry on an empirical study in the three wells oilfield reservoir type，the results of which are compared with the artificial neural network method，which show that the method is effective and feasible.Specific steps：firstly，neighborhood rough set is used to predigest the input attributes as a front pretreatment system and eliminate the redundant information，then put the remaining attributes as a Support Vector Machine（SVM） input variables，while the SVM as a rear system，it not only eliminates duplication of information between indicators，but also reduces the dimension.They make cooperation mutually，thus obtains the better measure results.

Key words: neighborhood, rough set, attributes reduction, support vector machine

摘要： 运用邻域粗糙集理论，对储层含油性的属性进行约简，并将约简后的属性作为支持向量机输入变量，对某油田的3口井油层类别进行实证研究，将结果与人工神经网络方法进行了比较,结果表明该方法是行之有效的方法。具体步骤为：先把邻域粗糙集作为前置系统对属性进行约减，剔除冗余信息，将剩余的属性作为支持向量机的输入变量。而支持向量机作为后置系统，不仅能消除指标之间信息重叠，而且可以降维。它们之间各司其责，相互配合从而得到好的评价结果。

关键词: 邻域, 粗糙集, 属性约简, 支持向量机