Online Hot Topic Detection from Web News Based on Bursty Term Identification

Abstract

With the increment in the volume of information, it’s almost impossible for people to assimilate all the news in time. A method to automatically detect hot topics from web news is strongly desired. Existing solutions take different perspectives ranging from identifying frequencies of terms to terms’ distribution or part-of-speech characteristics. However, most of them are either too simplistic or unfitting to the properties of hot topics. Therefore, this paper presents a hot topic detection approach based on bursty term identification. We propose a new bursty term identification approach which considers both frequency and topicality properties to detect the bursty terms and hot topics. A series of experiments have demonstrated that our proposed approach has good performance compared with baseline methods.