Dimensions overall score 5.0

GitHub Code Pulse

No public code linked for this paper yet.

• The paper proposes an ontology that models potentially toxic words in the Bulgarian language.method 95%
• A dataset of 4,384 manually annotated sentences from Bulgarian online forums was created, covering four categories: toxic language, medical terminology, non-toxic language, and terms related to minority communities.method 95%
• A BERT-based model trained for toxic language classification on the Bulgarian dataset achieves an F1 macro score of 0.89.result 95%
• The trained BERT model is directly applicable in a real environment and can be integrated into toxic content detection systems.technical 90%
• The presented approach aims to be more nuanced than current solutions to avoid inadvertently blocking valuable information like medical terms and text related to minority groups.method 85%
• The developed methodologies have potential applications across diverse online platforms and content moderation systems.market 80%
• Current solutions for toxic content detection often inadvertently block valuable information, including medical terms and text related to minority groups.limitation 75%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card