model reconciliation

The Model Reconciliation Problem (MRP) is a paradigm originated in the planning community as a method for explaining plans to human users. It has gained a lot of success due to the fact that it is rooted in the understanding of the importance of the so-called Theory of mind. This (normative) theory is used to explain some operations of the human mind and behavior in social and collaborative (or even adversarial) scenarios. In a nutshell, the Theory of Mind is the ability to attribute mental models to others while recognizing that these models may differ from one’s own. Now, in the context of MRP and planning, an AI agent and a human user typically have their own models that characterize the particular planning problem (e.g., fluents, actions, etc.). When these two models diverge, such that a valid plan in the agent’s model is invalid in the human’s model, the agent seeks to generate an explanation to reconcile these differences. The explanation is then used to update the human model, thus making the agent’s plan valid in the human’s (updated) model.

In my research, we have approached MRP from the perspective of knowledge representation and reasoning, where now the models of the agent and the human user are expressed in an appropriate logical formalism. By using logic, we have generalized MRP to problems beyond planning, so long as these problems can be expressed in some logic for which satisfiability of subsets can be decided.

selected publications

  1. ECAI
    PLEASE: Generating Personalized Explanations in Human-Aware Planning
    Stylianos Loukas Vasileiou , and William Yeoh
    In European Conference on Artificial Intelligence , 2023
  2. JAIR
    A Logic-based Explanation Generation Framework for Classical and Hybrid Planning Problems
    Stylianos Loukas Vasileiou , William Yeoh , Tran Cao Son , and 3 more authors
    Journal of Artificial Intelligence Research, 2022
  3. ICAPS
    VizXP: A Visualization Framework for Conveying Explanations to Users in Model Reconciliation Problems
    Ashwin Kumar , Stylianos Loukas Vasileiou , Melanie Bancilhon , and 2 more authors
    In International Conference on Automated Planning and Scheduling , 2022
  4. JELIA
    Model Reconciliation in Logic Programs
    Son Tran Cao , Van Nguyen , Stylianos Loukas Vasileiou , and 1 more author
    In European Conference on Logics in Artificial Intelligence , 2021
  5. AAAI
    On Exploiting Hitting Sets For Model Reconciliation
    Stylianos Loukas Vasileiou , Alessandro Previti , and William Yeoh
    In Association for the Advancement of Artificial Intelligence , 2021