Bridging the Divide: Promoting Serendipitous Discovery of Opposing Viewpoints with Visual Analytics in Social Media

Mahmood Jasim, Mumtaz Fatima, Sagarika Ramesh Sonni, Narges Mahyar

Room: 105

2023-10-22T22:00:00ZGMT-0600Change your timezone on the schedule page
Exemplar figure, described by caption below
Serendyze enables several features for social media post exploration including a search option to look for a specific word, a set of filters corresponding to representative pairs of keywords, filters for social media posts with For, Against, and Neutral alignments, three exploration metrics - Visit, Coverage, and Distribution, all social media posts, and suggested social media posts generated by the bias mitigation model that the readers may find interesting. This image was taken during P14’s exploration of social media posts.

hile social media promises open access to information, prior works suggest that it also plays a role as a catalyst for the social divide, which is often attributed to a shift towards algorithmic content curation based on users' digital footprints. To combat this issue, methods that support serendipity has received attention in recent years that aim to provide information beyond a user's viewpoint or preference. However, the utility of systems that promote serendipity in raising awareness of opposing viewpoints remains underexplored, especially in the political context. To that end, we conducted a study where we asked 14 participants to explore tweets about two politically charged topics --- gun control and immigration --- using an interaction-driven visual analytics tool that visualizes users' exploration patterns and provides serendipitous suggestions from opposing viewpoints. We found that as participants explored the tweets, they gradually became aware of opposing viewpoints and identified information they did not consider before which helped them gain knowledge about arguments from all sides. We also found while people were keen to use technology that promotes serendipity to cover more topical information, they do not necessarily trust the information found on social media. We hope that our work will motivate future researchers to investigate serendipitous aspects in visualizations to promote a more holistic exploration of various viewpoints.