17th International Conference on Social Computing, Behavioral-Cultural Modeling, & Prediction and Behavior Representation in Modeling and Simulation
September 18-20, 2024, Hybrid

More than 64 countries with varying levels of democracy will be hosting elections in 2024, including 10 of the largest, Bangladesh, Brazil, India, United States, Indonesia, Pakistan, Russia, and Mexico.

Along with ongoing active conflicts in Gaza, Sudan, and Ukraine, the risks of conflicts arising from climate, international and national financial, and government instability continue to increase.

The debates and elections of 2024 will have significant impacts on policies that will shape local and global outcomes over the next 10 years and beyond.

In this year’s SBP-BRiMS challenge problem, we ask participants to consider 2024 election topics and outcomes and address one or more of the following problems:

• How can new innovations in Computational Social Science (CSS) improve election outcome prediction?
• How can we effectively model, predict and influence voter turnout and voter access?
• Investigate the geographic and demographic factors that could affect voter turnout in a country's 2024 election. Use the data to map likely high and low turnout areas and propose strategies to increase participation.
• Election topics and debates
  • How can we effectively differentiate between Human and Bot stances in the media?
  • How does the analysis of topics change across scales (e.g., local, national, regional)?
• How do early election debates and outcomes influence later elections across regions?
  • Is there statistical significance between the election result(s) from one region(s) that affect another region(s)?
• How can exit polling be predicted topic stances in social media?

Participants may also submit other election-related challenge questions for SBP-BRiMS committee approval in a separate email to sbp-brims@andrew.cmu.edu

Course or thesis projects are welcome submissions with the approval of students’ professors or advisors.

Participants can use a variety of Computational Social Science (CSS) methodologies, including but not limited to:

• Large Language Models
• Machine Learning
• Modeling and Simulation
• Network Analysis

Datasets

The following datasets are publicly available and contain information relevant to the challenge. Participants may extend these datasets to obtain topical data and specific local data or choose their own from other sources such as social media.

Any unlisted datasets used in the challenge must be made publicly available.

• Reuters
• American National Election Studies
• Data.Gov
• Princeton University Library
• Voter Turnout Database
• University of South California Repo

There is no specific Reuters dataset that is recommended for the Challenge Question. Reuters did release a large corpus of data for R&D. Information on how to access these data are available here: Reuters data An older data set, Reuters-21578, is available here, old data. The data can also be found on Kaggle and other sites.

Evaluation Criteria

Each entry for the challenge problem should address one or more of the questions above.

All entries must have both a strong social theory, political theory or policy perspective, and a strong methodology perspective.