An Automatic e-news Article Content Extraction and Classification

This paper presents innovative automatic e-news article extraction and classification procedure. Most of the traditional methods need to analyse the layout of e-news websites. This paper proposed an efficient way to extract e-news using Rich Site Summary (RSS) feeds and Uniform Resource Locators (URLs) without analysing page layout. Newspaper python library is used for automatic e-news extraction process with enhanced features. Experimental results illustrate that the approach can extract e-news article content automatically and accurately. In classification module, this paper proposed ensemble classifier by combining three supervised learning algorithms. Ensemble classifier constructed using three individual classifiers. They are Support Vector Machine (SVM), Random Forest and Multinomial Naïve Bayes. These algorithms show higher accuracy than other individual classifier and the ensemble classifier improved the accuracy of the system. In this research, soft voting method is used to ensemble individual classifiers. It considers probabilities of individual classifiers. This research uses BBC e-news data set with 2225 news articles therefore research is limited to six main e-news categories. Such as political, sport, entertainment, business, tech and other. In here “other” category is separated using ensemble probabilities.