Tracking Elections: our experience during the presidential elections in Ecuador

TL;DR

This paper presents a methodology and tools for tracking election-related social media activity, analyzing Ecuador's 2017 presidential elections, and detecting political bots influencing online campaigns.

Contribution

It introduces a comprehensive approach for collecting, analyzing, and identifying political bots in social media during elections, with a case study on Ecuador's 2017 elections.

Findings

High presence of followers may be artificially inflated by bots.

Over 30,000 political bots were detected and analyzed.

Bots were mainly used for propaganda supporting or opposing candidates.

Abstract

The world's digital transformation has influenced not only the way we do business, but also the way we perform daily activities. In fact, the past Presidential elections in the United States as well as those in Great Britain (Brexit) and in Colombia (peace agreement referendum) are proof that social media play an important part in modern politics. In fact, this digital political field is filled by political movements and political candidates looking for popular support (number of followers), regular citizens' messages discussing social issues (trending topics flooding social media), or even political propaganda in favor or against politicians or political movements (advertisement). One of the issues with social media in this era is the presence of automatic accounts (bots) that artificially fill accounts with fake followers, create false trending topics, and share fake news or simply flood the net with propaganda. All this artificial information may influence people and sometimes may even censor people's real opinions undermining their freedom of speech. In this paper, we propose a methodology to track elections and a set of tools used to collect and analyze election data. In particular, this paper discusses our experiences during the Presidential Elections in Ecuador held in 2017. In fact, we show how all candidates prepared an online campaign in social media (Twitter) and how the political campaign altered a common follower rate subscription. We discuss that the high presence of followers during the period between the first and second round of elections may be altered by automatic accounts. Finally, we use bot detection systems and gathered more than 30,000 political motivated bots. In our data analysis, we show that these bots were mainly used for propaganda purposes in favor or against a particular candidate.