Minimax regret based elicitation of generalized additive utilities
This work addresses preference elicitation for decision-making in multiattribute domains, but it appears incremental as it builds on existing GAI and minimax regret frameworks.
The paper tackled the problem of eliciting generalized additively independent utilities using the minimax regret criterion, proposing new query types and strategies that exploit local structure for computational feasibility, enabling practical applications in preference-based configuration and product search.
We describe the semantic foundations for elicitation of generalized additively independent (GAI) utilities using the minimax regret criterion, and propose several new query types and strategies for this purpose. Computational feasibility is obtained by exploiting the local GAI structure in the model. Our results provide a practical approach for implementing preference-based constrained configuration optimization as well as effective search in multiattribute product databases.