Host: The Japanese Society for Artificial Intelligence
Name : The 37th Annual Conference of the Japanese Society for Artificial Intelligence
Number : 37
Location : [in Japanese]
Date : June 06, 2023 - June 09, 2023
During a sports match broadcast on TV or the Internet, many tweets about the match are posted on Twitter and it is possible to get rough progress of the match by reading these tweets. However, the purpose of tweets varies, and the content of tweets about a match ranges from those that contain a lot of information about the match to those that only express personal feelings, and thus it is not easy to instantly grasp the progress of the match from these tweets. In this study, we focus on soccer matches and work on building a system to generate live sports updates from tweets so that users can instantly grasp the progress of the match. We propose models based on a large pre-trained language model called T5 that incorporates a mechanism to control the number of updates and a mechanism to mitigate redundancy, and show their effectiveness through experiments.
