User interface. Map with gps coordinates related to social media posts
Architecture diagram

Inspiration

The ongoing conflict in Ukraine has become one of the most documented wars in history, with a massive influx of video, audio, and satellite data. With this wealth of information, we aim to support Ukraine by harnessing this data to reveal critical insights.

What It Does

Our project, "Operation Sentinel: Strategic Media & Sentiment Analysis," is designed to analyze foreign social media and news to identify top topic hierarchies, narratives, and entities with strong sentiment. The system collects data from diverse sources, including:

Telegram channels
VK.com
News websites +YouTube

It creates a structured dataset for each article, post, or video, capturing information such as:

Alert level
Topics
Narratives
Entities
Context
Timestamp
GPS coords
Emotions

This information is visualized geographically, displaying key insights on a map.

How We Built It

We leveraged several technologies and datasets, including:

OpenAI
Python
kepler.gl
Radiant
Primer dataset
Our custom dataset based on Telegram channels

Challenges We Ran Into

Processing large volumes of data poses challenges, especially with the increased time required for ETL processes and data collection.

Accomplishments We’re Proud Of

Created our own dataset from pro-Russian Telegram channels with over 1 million followers. Successfully performed ETL using the Primer dataset. Developed impactful visualizations using kepler.gl.