Projects / AI Innovation Challenge

Monitoring Human Rights Abuses Using AI

Project completed!


Monitoring Human Rights Abuses Through AI

Background

The rapid evolution of social media over the past decade has revolutionized global communication, generating vast amounts of data. This influx of data offers an unprecedented opportunity to monitor social phenomena in near real-time, including the ability to detect and respond to reported human rights abuses. Timely identification of such abuses can facilitate prompt action by governments and civil society to reduce harm. However, the challenge lies in efficiently processing and analyzing this data to extract meaningful insights.

Objective

The primary goal of this project was to develop a minimum viable product (MVP) for a real-time monitoring tool. This tool aimed to collect and analyze social media data to identify, categorize, and visualize potential human rights abuses. Partnering with Human Rights First, a leading international human rights organization, the project sought to build an application that could serve as an early warning mechanism for governments and advocacy groups.

Approach

To tackle the challenge, the project focused on:

  1. Data Collection: Extracting data from social media platforms such as Twitter, Reddit, Instagram, and Telegram.
  2. Language Modeling: Employing advanced Natural Language Processing (NLP) techniques, including ensemble models like BERT, to cluster, label, and compare text for identifying “ground truth” instances of human rights abuses.
  3. Categorization and Visualization: Creating a system capable of characterizing human rights abuses into well-defined classes and visualizing findings on a user-friendly dashboard.
  4. Real-Time Analysis: Developing a scalable back-end infrastructure to store, process, and analyze data in real-time, with confidence intervals and severity scoring for incidents.
Combating Human Rights using NLP - Source: Omdena

The final Dashboard MVP

Results and Impact

The project delivered a functional MVP of a human rights monitoring dashboard. Key outcomes included:

  • Automated Identification: Real-time detection of potential human rights abuses across multiple platforms.
  • Data Categorization: Effective clustering and labeling of incidents into predefined categories.
  • Visualization: A front-end dashboard showcasing trends and severity scores for easier interpretation and action.
  • Broader Impact: The tool empowers stakeholders, including governments and non-profits, to respond to human rights violations promptly, potentially reducing harm and saving lives.

Future Implications

The insights from this project could significantly influence future policies and practices. Governments could adopt such tools to proactively address human rights abuses, while researchers and advocacy groups could use the system to track trends and inform intervention strategies. Moreover, the project paves the way for further advancements in AI-driven monitoring systems, expanding their application to other social issues and global challenges.

This challenge is hosted with our friends at
Human Rights First


Thumbnail Image
Analyzing Air Quality in Gurugram Using Machine Learning
Thumbnail Image
[Nigerian Chapter] Improving Access to WASH Services for Displaced and Vulnerable Communities
Thumbnail Image
Identifying Motion-Based Actions to Reduce Carbon Footprint Using Motion Sensors

Become an Omdena Collaborator

media card
Visit the Omdena Collaborator Dashboard Learn More