基于特征倾向性的网页特征提取方法研究

Research on web page feature extraction method based on semantic orientation

网页包含的信息很丰富,传统的TFIDF公式很难满足内容过滤系统的要求。针对网页过滤技术中的特征选择方法存在的问题,加入语义信息,改进TFIDF公式,提出了一种比较适合网页过滤的特征选择方法。该方法综合考虑特征的长度、在网页中的位置信息,并且将情感色彩这种语义信息附加到特征上。实验结果表明,该方法在网页过滤系统中取得了较好的效果,尤其是实时内容过滤系统中,具有一定的实用价值。

Because the html page contains rich information, the traditional TFIDF formula is difiicult to meet the requirements ofcontent filtering systems. Some feature extraction methods for web filtering exist problems, semantic information is added, the TFIDF formula is improved and then a method of featttre extraction is proposed. It is more suitable for web filtering. This method considers the length of features, the location information of pages and the semantic orientation information of features. The experimental results show that the method is suitable for feature extraction in real-time web page filtering with some practical value.