A Very Efficient Approach to News Title and Content Extraction on the Web

摘要

We consider the problem of efficient and template-independent news extraction on the Web. The popular news extraction methods are based on visual information, and they can achieve good accuracy performance, but the computational efficiency is poor, because it is very time-consuming to render web page to obtain visual information. In this paper we propose an efficient and effective news extraction approach based on novel features. Our approach neither needs training nor needs visual information, so it is simple and very efficient. And it can extract news information from various news sites without using templates. In our experiments, the proposed approach achieves 99% accuracy over 5,671 news pages from 20 different news sites. And the efficiency is much faster than the baseline machine learning method using visual information.