MapSatisfyBench: Benchmarking Satisfaction-Aware Map Agents through Behavior-Grounded Implicit Decision Factors

The article introduces MapSatisfyBench, a new benchmark designed to evaluate satisfaction-aware map agents by addressing implicit decision factors crucial for user satisfaction. The proposed restore-identify-filter framework reconstructs user needs from behavior-chain evidence and identifies relevant implicit factors, facilitating a comprehensive evaluation of map agents beyond mere task completion. This benchmark is significant for AI practitioners as it shifts the focus towards enhancing user satisfaction in everyday interactions with map services, highlighting the need for agents that can proactively understand and respond to user needs.