This tool demonstrates a BiLSTM model's effectiveness in identifying harmful content across text and video platforms, implying potential for enhanced digital safety.
Key Points
The deployed model achieved a Recall of 0.9786, indicating high effectiveness in detecting harmful content.
User evaluations showed 95.45% rated the experience as Excellent, supporting the tool's strong acceptance in promoting safer interactions.
The deep learning model utilized a BiLSTM architecture with learned embeddings across various content categories, enhancing its multi-label classification ability.
Incorporating Whisper ASR enables detection of harmful speech in videos, highlighting versatile application in text and audiovisual contexts.