Did that happen? Predicting Social Media Posts that are Indicative of what happened in a scene: A case study of a TV show

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 7209–7214 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 7209 Did that happen? Predicting Social Media Posts that are Indicative of what happened in a scene: A case study of a TV show Anietie Andy 1, Reno Kriz4, Sharath Guntuku2, Derry Tanti Wijaya3, Chris Callison-Burch2 Penn Medicine 1, Computer and Information Science, University of Pennsylvania 2, Computer Science, Boston University 3, Human Language Technology Center of Excellence, Johns Hopkins University 4 Anietie.Andy@pennmedicine.upenn.edu, rkriz1@jh.edu, wijaya@bu.edu {ccb,sharathg }@cis.upenn.edu Abstract While popular Television (TV) shows are airing, some users interested in these shows publish social media posts about the show. Analyzing social media posts related to a TV show can be beneficial for gaining insights about what happened during scenes of the show. This is a challenging task partly because a significant number of social media posts associated with a TV show or event may not clearly describe what happened during the event. In this work, we do the following: (a) collect social media (Twitter) posts associated with a TV show, Game of Thrones, (b) identify the various scenes of the TV show, (c) for each scene, we annotate the social media posts (published during the time period of the scene) that are indicative of what transpired during the scene, and (d) we propose a method to predict the indicative social media posts in each scene. We show that for each of the identified scenes, with high AUC’s, our method can predict posts that are indicative of what happened in a scene from those that are not-indicative. Based on Twitter’s policy, we will make the Tweeter ID’s of the Twitter posts used for this work publicly available; also, we will make the annotations available. Keywords: Twitter, Game of Thrones, indicative, non-indicative 1. Introduction Prior works have shown that while popular televised events such as TV shows are airing, some individuals interested in these events publish social media posts associated with the events (Shen et al., 2013; Andy et al., 2017; Andy et al., 2019; Andy et al., 2020). Some of these posts capture user reactions to the various scenes associated with the events. In this work, using social media data associated with an episode of a TV show, Game of Thrones season 7 (GoT), we aim to identify the various scenes associated with the show and use a machine learning model to detect the posts that describe part/all of what happened during each scene. Similar to prior work (Panda et al., 2018), in this work, we define a scene as comprising of several shots focused on a particular idea/concept. Prior work identified and summarized scenes associated with a TV show by selecting a few representative so- cial media posts (5 posts) that were published during the timeline of the scenes (Andy et al., 2019). One of the challenges of this method is that some scenes are made up of several smaller sub-events and so selecting …

Content truncated to 500 words