Abstract:
This paper is aimed at news classification on basis of their headlines. Researchers have worked a lot for carrying out news classification at full text level but work in the domain of news headlines classification exists in very limited ratio, Therefore, after analyzing variety of existing news classification methodologies, a probabilistic framework is presented in this paper for classifying news headlines. News headlines classification process is divided into three modules, headlines pre-processing module, probability learning module, and news headlines classification module. Based on availability of variety of headlines, probabilistic framework is designed, which classifies each news headline to its pre-defined category by calculating its maximum probability in that category. Work has been performed using bag of words approach where each headline is split into words and each word is given a certain probability. Furthermore, it is shown that proposed system gives better accuracy results as compared to existing headline classification systems.
