A measurement framework and live report for the quality of an AI shopping assistant's product recommendations — graded relevance, weekly tracking, root-cause analysis, and judge calibration. Built on synthetic public data.
python evaluation data-quality search-relevance product-analytics streamlit ai-evaluation recommendation-quality machine-learning-evaluation llm-as-a-judge
-
Updated
Jun 10, 2026 - HTML