negmas-elicit

A library for preference elicitation during automated negotiations. This module was extracted from the negmas library to provide a focused, standalone package for elicitation capabilities.

Features

• Multiple Elicitation Strategies: Pandora-based elicitors, Value of Information (VOI) elicitors, and baseline implementations
• Query System: Flexible query types including range, rank, comparison, and marginal neutrality constraints
• User Modeling: Simulate users with different response behaviors and costs
• Expector Functions: Various strategies for handling uncertainty (mean, max, min, balanced, aspiring)
• Mechanism Integration: SAOElicitingMechanism for running negotiations with preference elicitation

Installation

From PyPI

pip install negmas-elicit

From Source

git clone https://github.com/yasserfarouk/negmas-elicit.git
cd negmas-elicit
pip install -e .

Quick Start

from negmas import MappingUtilityFunction
from negmas.outcomes import make_issue
from negmas_elicit import (
    User,
    PandoraElicitor,
    SAOElicitingMechanism,
)

# Define the negotiation issues
issues = [make_issue(10, "price"), make_issue(5, "quality")]

# Create a user with a known utility function
ufun = MappingUtilityFunction(
    mapping=lambda o: o[0] / 10 + o[1] / 5,
    issues=issues,
)
user = User(ufun=ufun, cost=0.1)

# Create an elicitor
elicitor = PandoraElicitor(user=user)

# Use in a mechanism or standalone elicitation
# ...

Available Elicitors

Baseline Elicitors

• DummyElicitor: No elicitation, uses prior beliefs
• FullKnowledgeElicitor: Assumes complete knowledge of user preferences

Pandora Elicitors

• PandoraElicitor: Standard Pandora's box approach
• OptimalIncrementalElicitor: Optimal incremental elicitation
• FastElicitor: Fast approximation
• MeanElicitor, BalancedElicitor, AspiringElicitor: Different expectation strategies
• OptimisticElicitor, PessimisticElicitor: Optimistic/pessimistic strategies

VOI Elicitors

• VOIElicitor: Value of Information based elicitation (OQA)
• VOIFastElicitor: Fast VOI approximation
• VOIOptimalElicitor: Optimal VOI strategy
• VOINoUncertaintyElicitor: VOI without uncertainty modeling

References

The elicitation algorithms implemented in this library are based on the following papers:

BibTeX

@inproceedings{baarslag2015optimal,
    title={Optimal incremental preference elicitation during negotiation},
    author={Baarslag, Tim and Gerding, Enrico H},
    booktitle={Proceedings of the 24th International Conference on Artificial Intelligence},
    pages={3--9},
    year={2015},
    organization={AAAI Press}
}

@inproceedings{baarslag2017value,
    title={The value of information in automated negotiation: A decision model for eliciting user preferences},
    author={Baarslag, Tim and Kaisers, Michael},
    booktitle={Proceedings of the 16th Conference on Autonomous Agents and MultiAgent Systems},
    pages={391--400},
    year={2017},
    organization={IFAAMAS}
}

@inproceedings{mohammad2018fastvoi,
    title={FastVOI: Efficient utility elicitation during negotiations},
    author={Mohammad, Yasser and Nakadai, Shinji},
    booktitle={International Conference on Principles and Practice of Multi-Agent Systems},
    pages={560--567},
    year={2018},
    organization={Springer}
}

@inproceedings{mohammad2019optimal,
    title={Optimal value of information based elicitation during negotiation},
    author={Mohammad, Yasser and Nakadai, Shinji},
    booktitle={Proceedings of the 18th International Conference on Autonomous Agents and MultiAgent Systems},
    pages={242--250},
    year={2019},
    organization={IFAAMAS}
}

Documentation

Full documentation is available at https://yasserfarouk.github.io/negmas-elicit/

Requirements

• Python 3.12+
• negmas >= 0.10.0

License

This project is licensed under the GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later). See the LICENSE file for details.

Citation

If you use this library in your research, please cite the negmas library:

@inproceedings{negmas2019,
  title={NegMAS: A Platform for Automated Negotiation},
  author={Mohammad, Yasser and Greenwald, Amy and Nakadai, Shinji},
  booktitle={International Conference on Principles and Practice of Multi-Agent Systems},
  pages={343--351},
  year={2019},
  organization={Springer}
}

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.