Introduction to the workshop and program

Panelists: Mr. Arvind Gupta Prof. Krithi Ramamritham, Prof. Tanel Tammet, Mr. Steve Newell,

Moderator: Prof. Huan Liu

Over the last decade, there has been an increase in the use of machine learning techniques for the prediction and analysis of elections. The recent trend has been the use of text mining of social media data to predict election outcomes. In this work, we assess the usefulness of employing census data to understand the voting patterns of South African voters during the 2016 local government elections at the electoral district level. We frame the problem as an unsupervised learning problem. Self-organising maps are utilised due to their visual nature and the resulting accessibility of information to stakeholders. The analysis shows that race, employment status, level of education, internet access and quality of service delivery in local government jurisdictions are key in determining which political party dominates that particular electoral district. These results could be used by the Electoral Commission of South Africa to prevent voter miss-information, especially for those voters who are in remote locations with little access to resources so as to ensure credible, free and fair elections in November 2021. Further, the results also present signals for socioeconomic reforms that can be beneficial for both the electorate and candidates.

This paper introduces VeriClaim, a computational fact-checking system that enables the automatic monitoring and verification of factual claims. VeriClaim is comprised of two computational modules: the claim-spotter, which first identifies check-worthy factual statements from swaths of online information, and the claim-checker, which verifies each statement using evidence retrieved from a multitude of knowledge resources. These two components are integrated and deployed in several ways, including a content-rich website and Chrome browser extension.

QVSR is an online survey platform for implementing interactive surveys that involve the application of the Quadratic Voting (QV) mechanism. QVSR was built in response to the shortcomings of Likert’s scale in eliciting true and relative preferences, minimizing polarization and addressing tyranny of the majority outcomes. The platform is robust and built for modern web browsers. It allows for integration into Qualtrics, deployment of QV or Likert surveys, analytics, data export and is intended to be open sourced. This platform aims to aid researchers in conducting QV surveys at scale, and with an intuitive, delightful user experience.

Social media has been gradually emerging at the forefront of election campaigning, and this transition has been further accelerated with the ongoing pandemic. In this extended abstract, we describe a mixed-methods approach to study paid election campaigning by contesting candidates, as well as parties, on Facebook. We have employed this approach to analyze online campaigning during two major Indian state legislative assembly elections - Bihar (November 2020) and West Bengal (May 2021). This analysis provides insights on the effectiveness of online campaigning strategies of contesting candidates and parties, while also making the discourse more accessible to the voters. We also compare this discourse with the tweets by the same users to study the differences between paid and unpaid online campaigning.

Many democracies in the world use district-based elections, where the region is divided into districts and there is a seat corresponding to each district in the governing body. In each district, several candidates representing different parties contest the election and residents of the district cast their votes. A candidate is declared the winner of the corresponding seat using a scoring rule on the votes, and the election result is understood in terms of the number of seats won by the candidates from different parties. This system is known to have a number of flaws, for example there is no guarantee that the election result truly reflect the popular support for different parties. The system may be gamed by different parties and authorities in different ways to produce a result that they prefer, and some 11 people’s opinions may get fully neglected and they find themselves unrepresented 12 and outside the power structure. It is important to identify the appropriate rules 13 for drawing districts, voting and scoring, so that the results satisfy various criteria 14 related to ethics and fairness. However, this requires exploring the huge range of 15 possible public opinions, which is difficult to do analytically. Here we propose 16 a framework for stochastic simulation of election results, which will allow all 17 stakeholders to explore the possible effects of different rules and enable them to 18 choose a more robust and ethical system. We also discuss how the simulation 19 models can be calibrated with available information using Machine Learning, so 20 that it can produce feasible results.

Panel 2: Panelists: Mr. Jack Cable, Dr. Jeremy Epstein, Prof. Sagar Samtani Moderator: Prof. Anita Nikolich