Live Semantic Sports Highlight Detection Based on Twitter

• Main Authors: Ching-Wei Lee, 李勁葦
• Other Authors: 徐宏民
• Format: Others
• Language: en_US
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/33474976272580076837

碩士 === 國立臺灣大學 === 資訊網路與多媒體研究所 === 99 === Twitter, "SMS of the Internet" [3], a world-wide social networking and microblogging website. The real-time and conversational properties have attracted the attention of scholars and researchers in recent years. Twitter experienced rapid growth and its usage spikes during prominent events, especially in sports. For example, “a record was set during the 2010 FIFA World Cup when fans wrote 2,940 tweets per second in the thirty-second period after Japan scored against Cameroon on June 14, 2010. The record was broken again when 3,085 tweets per second were posted after the Los Angeles Lakers'' victory in the 2010 NBA Finals on June 17, 2010,[36] and then again at the close of Japan''s victory over Denmark in the World Cup when users published 3,283 tweets per second.[37] The current record was set during the 2011 FIFA Women''s World Cup Final between Japan and the United States, when 7,196 tweets per second were published.” [1] These results of Twitter usage properly show our motivation of this study. In this paper, we detect the burst of Twitter usage which we called highlight while sports games broadcasting and acquire the semantic meaning of the highlights. The highlights precisely represent the highly-discussed topics which Twitter users focus on during the game time. We present the observations of the phenomenon that what Twitter users are interested in while watching sports games. Finally, we compare the highlights with traditional sports events and analyze in each sports domain.