Abstract:
|
Before the arrival of modern information and communication technology, it was not easy to capture people's consuming and company-rating preferences; however, the prevalence of social-networking websites provides opportunities to capture those trends in order to predict social and economic changes. Latent topic analyses, such as LDA, are good at discovering topics from large sets of textual data. High-quality information is typically derived through the devising of patterns and trends through topic models, a statistical learning method. While traditional topic models perform very well for news articles and journals archives, for those documents of a short length (e.g. less than 140 characters for Micro-blog), LDA lacks the ability to find fine grain topics. In this paper, we propose variations of topic models that unveil the evolution of topics over time on Chinese micro-blog (Weibo). Approaches to resolve and eliminate the disturbance of randomness are attempted to generate more stable topic distributions. Methods for topic evolution analysis, such as TM-LDA, are employed to measure the strength and variability of topics.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.