Machine Learning Technique for Early Detection of Social Unrest Activities in Sri Lanka Using Twitter Data

Abstract

"Protests hold a long-standing legacy for creating shifts within societies and cultures. Throughout history, these demonstrations have often ignited profound changes in political systems. While quantifying these losses monetarily is challenging, their impact on businesses is profound, including disruptions to manufacturing and services. Sri Lanka is a country that is home to various manufacturing industries. In 2021, Sri Lanka exported a total of USD 14.1 Billion worth of goods and services that accumulated to 16.58% of Sri Lankan GDP. Sri Lanka is also a nation that has experienced a notable series of prominent demonstrations in the recent past. With its economy leaning heavily on exports, any protest capable of immobilizing a key manufacturing hub due to widespread protests and resultant curfews could markedly undermine the overall manufacturing productivity. Hence enabling businesses to receive advance alerts about potential protests that could impact manufacturing provides an opportunity for these businesses to implement pre-emptive measures and strategies to mitigate the ensuing disruptions. While prior research has explored many aspects of protests on a global scale including early detection, a critical void remained in investigating protest and its early detection using social media within the specific context of Sri Lanka. This research was undertaken to bridge this lack through a comprehensive analysis of social media data that could help Sri Lankan businesses to confront the effects that rises from unexpected protests. Hence, this study conducted an in-depth analysis of Twitter data from the 2022 Sri Lankan Protests, integrating keywords, location mentions, and sentiment analysis. Despite the limitations posed by a limited dataset, this study successfully uncovered that certain Twitter keywords and the negative sentiment expressed in Tweets do exert an influence on the incidence of protests in Sri Lanka, to a certain degree. "