Iterated Update: A Preference-Based Approach

Title: Iterated Update: A Preference-Based Approach

Abstract: This research article explores an approach to iterated update, a process where an agent receives a sequence of formulas representing successive states of the world. The agent uses an abstract distance or ranking function to determine the most likely outcome, making assumptions of inertia and ranking based on the likelihood of histories. The article provides a characterization of operators that correspond to such reasoning, including soundness and completeness properties. The study emphasizes the relevance of epistemic states in iterated update and provides a full characterization of these operators.

Main Research Question: How can an agent's knowledge about an evolving situation be updated using an abstract distance or ranking function?

Methodology: The study uses a preference-based approach to iterated update, making assumptions of inertia and ranking based on the likelihood of histories. The article characterizes operators that correspond to such reasoning, providing a set of conditions for the operator and a method to construct a ranking from the operator.

Results: The research provides a complete characterization of operators based on history ranking, including soundness and completeness properties.

Implications: The study's findings have implications for the field of artificial intelligence, as they provide a new approach to iterated update that can be applied in various scenarios involving evolving situations. The approach also contributes to the ongoing debate on the distinction between belief revision and update, offering an ontological perspective on the two processes.

Link to Article: https://arxiv.org/abs/0202026v1 Authors: arXiv ID: 0202026v1