Nomination: Best AI-Driven Solution
Veritas Watch – A system for identifying hate speech in the information space of hostile propaganda
Problem
Hostile propaganda is a powerful tool of psychological warfare against Ukraine, creating a multi-layered system of manipulation. Main challenges:
- Systematic dissemination of anti-Ukrainian narratives: propaganda discredits national values and the right to self-determination.
- Hidden forms of hate speech: sophisticated linguistic techniques disguise aggressive messages as neutral.
- Justifying armed aggression: creating false arguments to legitimize the occupation.
- Technical difficulty of identification: manual analysis is slow and automatic tools need to be improved.
This problem requires a systematic approach to identify and counteract.
Solution
Veritas Watch is an innovative NLP system for automatic analysis of media content. Its goal is to find and classify dangerous propaganda statements in five areas:
- Objections.
- Justification
- Appeals
- Incitement to hatred
- Humiliation of the dignity of Ukrainians
The system is based on machine learning technologies, in particular the BERT model, which ensures accuracy and deep context analysis.
Read the full analytical report with other solutions and the following sections of this solution:
- Key advantages
- Technologies
- Roadmap
- Expected result
- Conclusions