摘要：With the rapid development of Internet, information provided by the Internet has shown explosive growth. In the face of massive and constantly updated information on the Internet, how the user can fast access to more valuable and more information has become one of the hot spots. The time of Web Page update appears to be erratic, so forecasting the update time of news reports is even more difficult. From the view of application, we can use mathematical models to maximize the approximation of variation, although it cannot be completely accurate. So is the predicting the update time of news which helps in improving the news crawler’s scheduling policy. In this paper, we proposed a combined predict algorithm for news update. In order to predict the update time of news, firstly, we applied the Exponential Smoothing method to our dataset, and we also have selected the optimal parameters. Secondly, we leveraged the Naive Bayes Model for prediction. Finally, we combined two methods for Combination Forecasting, as well as made a compare with former methods. Through the experiments on Sohu News, we show that Combination Forecasting method outperforms other methods while estimating localized rate of updates.
关键词：Exponential Smoothing Method;Naive Bayes Model;Combination Forecasting;News Update Time
