Abstract
Although an increasing of people is getting their news from the Internet, news web sites do not provide sufficient categorization of the news. Currently, people have a hard time to go through news after news to find what is most relevant to their interests. This study presents a novel categorization algorithm for automatic categorization of news articles. Based on a large corpus of news titles and categories assigned by human experts, the algorithm dissects an in-coming news article and decides the most likely category. The algorithm relies on keywords that are most likely to appear in news titles of a particular category to make the decision. Experiments indicated that the positions of words are related to the category and this algorithm categorizes the daily news automatically and efficiently.