基于语义倾向性的文本过滤研究

Research of text filtering based on semantic orientation

以往文本过滤的研究主要集中于主题过滤,然而随着网络的发展,倾向性文本过滤在网络信息安全方面的作用越来越大。在语义倾向性分析中,若忽略关联词和修饰词则有可能导致对极性词的倾向或强度判断失误。针对这一问题,提出了一种新的语义倾向性识别算法,用于对潜在极性词进行倾向性识别,并应用到文本过滤方面。实验表明此方法具有较高的准确率和召回率。

The previous researches of text filtering mainly focused on the theme filtering. However, with the development of the Internet, the text filtering based on orientation is playing an increasingly important role in the network security. In the analysis of orientation, if the associated and qualifier words were neglected, the polarity and strength of the words might be misjudged. In order to solve the problem, this paper proposed a new algorithm to recognize the potential polarity words and applied it to the text filtering. The experiment shows that this method has higher accuracy and recall rate.