A Study of Predicting Popular Hashtags on Twitter by Combining the Role Features Information

• Main Authors: Ming-Han Tsai, 蔡明翰
• Other Authors: Yi-Leh Wu
• Format: Others
• Language: en_US
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/zuhaax

碩士 === 國立臺灣科技大學 === 資訊工程系 === 105 === With ubiquitous networks and mobile devices, people can easily spread and access information on the Internet at any time. But information on the Internet is often too messy, causing us to spend too much time on the absorption of trivial messages, yet too late to know or even missed the important messages. For this reason, we hope to know beforehand the outbreak of news in order to take early action. This kind of problem has many related applications in life, such as recommendation system, content filtering, disease prevention, marketing, etc. In recent years, more and more people have been involved in research in this field. In this paper, we propose a new method based on the model designed by Wang et al. and further combine the role features. We predict popular hashtags on a Twitter dataset that contains 595,460 users and 14,607 hashtags with final size no less than 50. The experiment results show that the role features information is helpful for the cascade prediction problems.