Topic-Level Bursty Study for Bursty Topic Detection in Microblogs
Microblogging services,such as Twitter and Sina Weibo,have gained tremendous popularity in recent years.The huge amount of user-generated information is spread on microblogs.Such user-generated contents are a mixture of different bursty topics(e.g.,breaking news)and general topics(e.g.,user interests).However,it is challenging to discriminate between them due to the extremely diverse and noisy user-generated text.In this paper,we introduce a novel topic model to detect bursty topics from microblogs.Our model is based on an observation that different topics usually exhibit different bursty levels at a certain time.We propose to utilize the topic-level burstiness to differentiate bursty topics and non-bursty topics and particularly different bursty topics.Extensive experiments on a Sina Weibo Dataset show that our approach outperforms the baselines and the state-of-the-art method.
Sina Weibo Bursty topic detection Topic model Hypothesis testing
Yakun Wang Zhongbao Zhang Sen Su Muhammad Azam Zia
Beijing University of Posts and Telecommunications,Beijing,China Beijing University of Posts and Telecommunications,Beijing,China;University of Agriculture Faisalaba
国际会议
澳门
英文
97-109
2019-04-14（万方平台首次上网日期，不代表论文的发表时间）