基于柔性匹配的中文文本特征提取方法

针对含有变形关键词的不良信息过滤问题，提出一种基于柔性匹配的中文文本特征信息提取方法。该方法采用柔性匹配技术识别和提取变形关键词，改进向量空间模型中特征项权重的计算方法，对具有变形形式的关键词赋予较高权重，从而提高特征信息的提取效率。实验结果表明，该方法可在保证过滤准确率的前提下，获得较高的召回率。

Feature Extraction Method in Chinese Text Based on Flexible Matching

Aiming at the problem of filtering malicious information which contains transformed keyword, this paper presents a feature extraction method in Chinese text based on flexible matching. The method adopts flexible matching technology to identify transformed keyword, improves the computational method of feature term weight in Vector Space Model(VSM). The keyword which has transmutative form is endowed high weight to enhance extraction efficiency for feature information. Experimental result shows that the method of feature information extraction for filtering has high recall in the condition of ensuring precision.