{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/716e5d080d844b62a072c09648cca338\" frameborder=\"0\" width=\"1660\" height=\"1245\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1245,"width":1660,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1245,"thumbnail_width":1660,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/716e5d080d844b62a072c09648cca338-2542cb3084606b11.gif","duration":595.328,"title":"Music Mate Finder 2.0 AI Recommender Demo","description":"Hi, I am Julie, and I presented my Music Mate Finder 2.0, an AI conversational music recommender built on a Scorpion scoring system. It has four layers, Streamlit UI, an agent orchestrator, a specialized model that converts your input into a structured profile, and a REG engine that searches song descriptions by meaning using vector embeddings. For clear requests like upbeat and happy, confidence is around 0.9, and for vague input like late night drive, it still returns good matches. I noted that very low confidence drops to about 0.4 when you say just music, and a future improvement is to ask for more detail then. No action was explicitly requested from viewers."}