Elections Watch

Citizen participation is generally agreed to be an essential ingredient of a healthy democracy[^1] and has been recognized as a UN sustainable development goal [^2]. Furthermore, the availability of data to track these goals has been identified as key ingredient in effective implementation[^3].

This project aims to improve improve citizen participation in the political process, especially for populations in the developing world (using Nigeria as an example). It leverages data from twitter in addressing this goal through the following modules.

  • Political Event Summary: Providing a longitudinal summary of key events that drove political discourse over the period of an election. In this work we provide an interactive visualizations (time series visualizations, word cloud visualizations) of key events over the period of 6 months of the 2019 election. The resulting catalog of events in turn are reflective of issues (sometimes related to justice and corruption) that require attention, and can serve as artifacts for communal memory on these issues.

  • Citizen and Candidate Participation: Provide an analysis of how various candidates (and citizens) utilized in twitter. This includes visualizations of tweets authored by candidates, replies to candidates and candidate replies.

  • Visualizations of Community Structure. Leveraging geotags from the twitter api, we demonstrate the existence of not only local communities by also global diaspora communities and their participation in election discourse.

Taken together, these modules represent a first step in understanding political discourse, and making the political institution more transparent, especially for developing regions for which the infrastructure or resources for more precise measurements are not available.

Data Collection and Application Development Process

  • Data for the project was collected using python scripts that wrote tweets from the Twitter v2 (previously v1) streaming api (matching certain criteria) to BigQuery. Criteria was based on a list of keywords (candidate names, known political hashtags, local political institutions) and locations (bounding box corresponding to Nigeria).

  • BigQuery queries were then used to generate aggregate datasets used for visualizations/analysis and training machine learning models (political text classification models to label political text and multi class classification models to label general discourse). The results from analysis were then written to json files that are visualized by the app.

Overall, the dataset for the 2019 elections contain 25.2 million tweets and retweets, 12.6 million original tweets 8.6 million geotagged tweets and 3.6 million tweets labelled (using an ML model trained on local political text) as political.

Next Steps

  • 2023 Elections: The current web application is based on tweets collected, pertaining to the 2019 election. Current work is focused on collecting data related to the 2023 election and extended analysis for integration in the Election Watch app.
  • Interaction Gamification: Implementing gamification mechanics to enable citizens better engage with results from this analysis. This is structured into two areas:
    • Sharing: e.g., providing the ability to share speicific insights from the analysis with explanations/evidence
    • Individual reflection: allowing individuals to login and analyze/reflect on their own political participation tweet data over time, organized by some leaderboard.

Note: In this project, we note that online demographics (people who self elect to use Twitter) may not be fully representative of the entire target population; the reader is encouraged to consider the results are best applicable to this sub-population.

[^1]: Citizen Participation and Democracy, Karen Bullock. https://link.springer.com/chapter/10.1057/9781137269331_2 [^2]: Sustainable Development Goal 16: Peace, justice and strong institutions, United Nations. https://sdgs.un.org/goals/goal16 [^3]: Participation, Consultation and Engagement: Critical Elements for an Effective Implementation of the 2030 Agenda. Amina J. Mohammed. https://www.un.org/en/chronicle/article/participation-consultation-and-engagement-critical-elements-effective-implementation-2030-agenda

Built With

Share this project: