Definition
Mean Average Precision (MAP) is a rank-aware metric that evaluates retrieval quality by averaging the precision at each relevant document's position across multiple queries. While it effectively penalizes RAG systems that bury relevant context, it relies on binary relevance and lacks the nuance of graded relevance found in metrics like NDCG.
Evaluates the ranking order of search results, not the statistical 'Maximum A Posteriori' probability.
"A scavenger hunt leaderboard where participants receive higher scores for finding the required items at the very beginning of their search path rather than at the end."
- Precision@K(Component)
- NDCG (Normalized Discounted Cumulative Gain)(Alternative Metric)
- Recall(Prerequisite)
- Ranked Retrieval(System Context)
Conceptual Overview
Mean Average Precision (MAP) is a rank-aware metric that evaluates retrieval quality by averaging the precision at each relevant document's position across multiple queries. While it effectively penalizes RAG systems that bury relevant context, it relies on binary relevance and lacks the nuance of graded relevance found in metrics like NDCG.
Disambiguation
Evaluates the ranking order of search results, not the statistical 'Maximum A Posteriori' probability.
Visual Analog
A scavenger hunt leaderboard where participants receive higher scores for finding the required items at the very beginning of their search path rather than at the end.