基于包装器模型的文本信息抽取

在分析基于标志和基于文本模式两类算法的基础上，提出了一种新的包装器归纳学习算法。新算法综合上述两类算法的优点，不但能利用页面的标志信息进行信息定位，而且能利用文本的模式信息来进行信息抽取和对抽取结果进行必要的过滤。实验结果表明，新算法具有较高的信息抽取精度与信息表达能力。

Text information extraction based on wrapper model

A new wrapper induction algorithm was proposed for text information extraction after analyzing two types of algorithms based on landmark and text pattern. The new algorithm can take the advantage of above-mentioned two algorithms. It can locate the information based on the landmark information of Web pages, and can use the text pattern to extract and filter large quantity of Web text. Experiment results show that the new method achieves higher accuracy and expressiveness of information extraction.