Defenders

We are working to fix the lack of understanding of communication between Russia and Ukraine through a social media platforms popular in both countries, satellite imaging, and news articles.

Comment

Inspiration

The lack of understanding surrounding communication efforts between Russia and Ukraine on Telegram highlights the urgent need to analyze communication patterns within war-torn regions. This project aims to understand anti-war sentiment and how information flows, potentially aiding in humanitarian and peace-building initiatives.

What it does

This project envisions a tool that:

Leverages NLP techniques to summarize and analyze textual data from sources like newspapers and potentially Telegram (if access is possible). Identifies key topics and overall sentiment trends within the collected data. Translates Russian sources to facilitate analysis.

How we built it

Dataset: We initially focused on Russian newspaper data. Ideally, we would expand our data sources to include Telegram discussions. NLP Tools: We would explore tools like Primer (if accessible), or open-source alternatives such as Hugging Face Transformers, NLTK, or spaCy. Implementation: We would develop scripts or a small application to process data, perform summarization, topic modeling, and sentiment analysis.

Challenges we ran into

Data Access: Reliable, uncensored Telegram data might be difficult to obtain. State Control: Russian newspapers are heavily influenced by the state, potentially skewing the sentiment analysis. Tool Complexity: Tools like Primer may have a learning curve and require specialized knowledge to use effectively.

Accomplishments that we're proud of

Problem Formulation: We clearly defined a relevant problem and articulated how NLP could provide insights. Solution Design: We outlined a potential technical approach even with limited data access.

What we learned

Censorship's Impact: The project underscored the challenges of analyzing sentiment and communication within highly controlled media environments. NLP's Power: We gained deeper appreciation for the potential of NLP tools in extracting meaning from complex text data.

What's next for Defenders

Secure Telegram Data: Explore ethical and secure ways to access relevant Telegram communication (if feasible). Open-Source Focus: Build our solution using accessible open-source NLP tools, making it more replicable. Collaboration: Seek partnerships with researchers or humanitarian organizations with interest in this domain.

Built With

  • claude
  • dedrone
  • primer
  • skyfi
  • vannevar
Share this project:

Updates