Download PDFOpen PDF in browserConstrained Adaptive Negotiation Agent of ODRL Offers Under Incomplete and Asymmetric Information12 pages•Published: October 28, 2026AbstractThe Open Digital Rights Language (ODRL) has emerged as a key policy language for governing data usage in data spaces and decentralized digital ecosystems. While ODRL enables machine-readable representation and enforcement of permissions, prohibitions, and obligations, automated negotiation under ODRL constraints remains insufficiently addressed. Autonomous agents should uphold these constraints on top of negotiating heterogeneous policy terms under incomplete information. This paper presents a policy-aware negotiation agent that operates directly on ODRL policy constraints while ensuring the semantic interpretability of negotiated offers. The proposed framework combines per-issue Bayesian belief models capable of handling mixed-type policy constraints, strict enforcement of non-negotiable constraints, and adaptation mechanisms that guide acceptance decisions and counter-offer generation. Unlike utility-driven or reinforcement learning approaches, the proposed agent enforces non-negotiable constraints by construction and restricts exploration to the feasible policy space. We position this framework as a practical step toward regulating autonomous agents with ODRL in open systems, such as data spaces, and report implementation insights from translating ODRL constraint semantics into executable negotiation logic.Keyphrases: automated negotiation, bayesian learning, data spaces, odrl, policy aware ai, policy enforcement In: Andrea Cimmino Arriaga, Beatriz Esteves, Nicoletta Fornara and Víctor Rodríguez Doncel (editors). Proceedings of 2nd Workshop ODRL and Beyond: Practical Applications and challenges for poLicy-base access and usage control, OPAL 2026, vol 113, pages 16-27.
|
