Computational Analysis of Bot Activity in the Asia-Pacific: A Comparative Study of Four National Elections

Abstract

Bot-driven electoral disinformation represents a major threat to democracies worldwide. Extant scholarship, however, tends to concentrate around Western contexts. This paper undertakes a comparative computational analysis of bot activity during four recent elections in the Asia-Pacific. Through a systematic, multi-level comparison of bot activity, we contribute novel insights about shared and distinct computational features of the disinformation landscapes within a significant yet understudied geopolitical region. Across case studies in Indonesia, the Philippines, Singapore, and Taiwan, we find non-negligible levels of bot activity: bots engage in higher levels of tweet production; interact with humans especially through mentions; tend to occupy denser and more isolated communities; use simpler and abusive language; and share partisan, irrelevant, or conspiratorial content. We conclude with implications for deepening and utilizing the analysis presented here as well as future directions for further cross-national work.