Fair dynamic pricing and certified-fair hotel room recommendations — combating overtourism overpricing with game-theoretic AI.
1University of Southampton 2Technical University of Crete 3Artificial Intelligence Research Institute (IIIA, CSIC)
Three complementary works addressing fair hotel pricing — from a workshop sketch to a full certified-fair recommendations platform.
The original workshop paper that introduced the core idea: a game-theoretic approach to fair hotel room pricing using cooperative game theory and the Owen value. Outlines a policy sketch where room prices reflect each hotel's relative power based on supply and demand, ensuring profit margins proportional to quality. Proposes a novel graph-based representation and presents initial experimental findings.
Introduces a game-theoretic dynamic pricing policy using Owen values to determine fair room prices based on supply and demand. Models the hotel room market as a cooperative game (the Dynamic Hotel-Rooms Game) and computes fair profit margins that reflect each room's "power" in the platform. Also introduces the Exposure-Owen Ratio to ensure fair provider visibility. Evaluated on real-world datasets from Athens, Barcelona, and Rome.
Extends FairPlay into a full recommendations platform with certified-fair prices. The framework is modular — it can incorporate any fairness notion as a pluggable fair-base pricing policy. It introduces an incentivisation mechanism (Theorem 1) proving that hotels are weakly better-off adopting the fair price. Hotels retain autonomy over their pricing while the platform adjusts exposition opportunities to promote fairness. Explainable by design.
Overtourism fuels overpricing, harming communities and tourists alike.
Popular destinations face overpopulation and profiteering. Hotels commonly quadruple room rates in high-demand periods. Existing regulations — tourist taxes, access fees, visitor limits — have shown limited effectiveness.
UNWTO Article 7.1 establishes equal access to tourist destinations as a fundamental right. Uncontrolled overpricing makes popular destinations inaccessible to a large portion of the populace.
Hotels account for ~80% of guest nights in EU destinations and ~75% of guest night growth in the top-10 EU cities, making the hotel industry the pivotal accommodation player.
While fair dynamic pricing algorithms exist, hotel owners are reluctant to relinquish pricing control to an algorithm. Appropriate incentives must be in place for voluntary adoption.
"The prospect of direct and personal access to the discovery and enjoyment of the planet's resources constitutes a right equally open to all the world's inhabitants."
— UNWTO Global Code of Ethics for Tourism, Article 7.1Download the papers and access the code.
The workshop paper introducing the core idea: a cooperative game-theoretic approach to fair hotel room pricing using the Owen value, with a novel graph-based representation.
Full paper with complete technical details, the Exposure-Owen Ratio for provider fairness, and systematic evaluation on real-world hotel data from Athens, Barcelona, and Rome.
A game-theoretic dynamic pricing policy using Owen values for multi-sided fairness.
FairPlay models the room pricing problem as a cooperative game (DHRG). Rooms are players, hotels are coalitions. A graph captures dependencies via three edge types: room-to-type (demand within a hotel), internal-type-to-type (cross-room-type demand), and external-type-to-type (same type across hotels).
cδ = uδ · (1 + Owδ)The final price increases linearly with the room's Owen value (power), ensuring prices reflect actual supply and demand rather than arbitrary markups.
FairPlay ensures fairness for both customers and providers simultaneously:
Customer-side: Prices increase linearly with room power — no profiteering.
Provider margins: Profit proportional to room's Owen value power.
Fair exposure: Exposure-Owen Ratio ensures proportional visibility.
Quality rewards: Higher-quality hotels earn larger profit margins.
A modular, certified-fair recommendations framework that incentivises fair pricing.
Guest submits check-in date, duration, room features, and budget to the platform.
Platform computes the fair price (via fair-base policy) alongside the hotel's own price for each room.
Exposition adjusted to incentivise fair pricing. Certified-fair rooms are promoted to the visitor.
Hotels overpricing above the fair margin get reduced exposition by a lowering factor λ, proportional to the gap between their price and the fair price.
λ = (1 + αfair) / (1 + αhotel)Pr(exposed; q) = min{λ, 1} · Pr(CF; q)Providers are in expectation (weakly) better-off when using the fair-base pricing policy.
The fair price is the dominant strategy — non-CF hotels never outperform.
Any fairness notion: Modular — swap the fair-base policy to any definition.
Provider autonomy: Hotels retain freedom to set their own prices.
Explainable by design: Transparent, traceable decisions — no black boxes.
Non-CF not starved: Every hotel always has Pr > 0 to be exposed.
Across both papers: real-world data (Athens, Barcelona, Rome) and synthetic simulations (48 hotels, 30 days).
FairPlay⁺ hotels achieve the greatest cumulative income — fair pricing is consistently the most profitable strategy for providers.
INCOMECertified-fair hotels exhibit the highest daily occupancy rates, confirming visitor preference for fairly-priced rooms.
OCCUPANCYFairPlay⁺ dramatically improves access for low-budget visitors, serving all budget profiles where static policies fail.
SATISFACTIONOn real-world data, static policies overprice rooms in 20 out of 30 days compared to FairPlay's demand-responsive pricing. FairPlay detects quality differences, rewarding high-demand hotels with larger margins while penalising low-quality providers — preventing profiteering even in monopoly scenarios. Provider exposure matches room power with >85% similarity across all datasets.
In 48-hotel simulations with up to 750 queries/day over 30 days, FairPlay⁺ hotels consistently outperform static +17%, +40%, and +50% policies. The advantage is strongest for low-budget visitors (≤45 MU, ≤65 MU), where FairPlay⁺ serves significantly more visitors — directly supporting the UN right to tourism.
Both works are designed with ethical AI principles at their core.
Supports UNWTO Article 7.1 by ensuring equal access to destinations for all budget profiles. Low-budget visitors find accommodation that would be inaccessible without fair pricing.
Hotels retain full freedom to set their own prices. The platform incentivises fair pricing through adjusted exposition rather than enforcing it — participation is voluntary.
Explainable-by-design decisions with no opaque algorithms. Users can trace how popularity, room features, and demand influence prices and exposition.
Balances visitor affordability with fair provider revenue and exposition opportunities. Neither side is disadvantaged — a balanced and ethical ecosystem.
@inproceedings{streviniotis2023fairplay,
title = {A Fair Dynamic Pricing Policy for the
Hotel Industry},
author = {Streviniotis, Errikos and Georgara, Athina
and Bistaffa, Filippo and
Chalkiadakis, Georgios},
booktitle = {5th Games, Agents, and Incentives Workshop
(GAIW 2023), held at the 21st International
Conference on Autonomous Agents and
Multiagent Systems (AAMAS)},
year = {2023},
address = {London, England}
}@inproceedings{streviniotis2024fairplay,
title = {FairPlay: A Multi-Sided Fair Dynamic
Pricing Policy for Hotels},
author = {Streviniotis, Errikos and Georgara, Athina
and Bistaffa, Filippo and
Chalkiadakis, Georgios},
booktitle = {Proceedings of the 38th AAAI Conference
on Artificial Intelligence (AAAI-24)},
pages = {22368--22376},
year = {2024}
}@inproceedings{georgara2026fairplayplus,
title = {FairPlay+: A Certified-Fair Hotel Rooms
Recommendations Framework},
author = {Georgara, Athina and Streviniotis, Errikos
and Bistaffa, Filippo and
Rodriguez-Aguilar, Juan A. and
Ramchurn, Sarvapali D. and
Chalkiadakis, Georgios},
booktitle = {Second Conference of the International
Association for Safe and Ethical
Artificial Intelligence (IASEAI'26)},
year = {2026}
}