Florence Dupin de Saint-Cyr

Penalty logic and its link with Dempster-Shafer theory

Penalty logic, introduced by Pinkas [17], associates to each formula of a knowledge base the price to pay if this formula is violated. Penalties may be used as a criterion for selecting preferred consistent subsets in an inconsistent knowledge base, thus inducing a non-monotonic inference relation. A precise formalization and the main properties of penalty logic and of its associated nonmonotonic inference relation are given in the first part. We also show that penalty logic and Dempster-Shafer theory are related, especially in the infinitesimal case.

Belief extrapolation (or how to reason about observations and unpredicted change)

We give a logical framework for reasoning with observations at different time points. We call belief extrapolation the process of completing initial belief sets stemming from observations by assuming minimal change. We give a general semantics and we propose several extrapolation operators. We study some key properties verified by these operators and we address computational issues. We study in detail the position of belief extrapolation with respect to revision and update: in particular, belief extrapolation is shown to be a specific form of time-stamped belief revision. Several related lines of work are positioned with respect to belief extrapolation.

Change in argumentation systems: exploring the interest of removing an argument

This article studies a specific kind of change in an argumentation system: the removal of an argument and its interactions. We illustrate this operation in a legal context and we establish the conditions to obtain some desirable properties when removing an argument.

Enforcement in Argumentation Is a Kind of Update

In the literature, enforcement consists in changing an argumentation system in order to force it to accept a given set of arguments. In this paper, we extend this notion by allowing incomplete information about the initial argumentation system. Generalized enforcement is an operation that maps a propositional formula describing a system and a propositional formula that describes a goal, to a new formula describing the possible resulting systems. This is done under some constraints about the allowed changes. We give a set of postulates restraining the class of enforcement operators and provide a representation theorem linking them to a family of proximity relations on argumentation systems.

Updating, Transition Constraints and Possibilistic Markov Chains

Possibility theory is applied to the updating problem in a knowledge base that describes the state of an evolving system. The system evolution is described by a possibilistic Markov chain whose agreement with the axioms of updating is examined. Then it is explained how to recover a possibilistic Markov chain from a set of transition constraints, on the basis of a specificity ordering.

A Comparative Study of Six Formal Models of Causal Ascription

Ascribing causality amounts to determining what elements in a sequence of reported facts can be related in a causal way, on the basis of some knowledge about the course of the world. The paper offers a comparison of a large span of formal models (based on structural equations, non-monotonic consequence relations, trajectory preference relations, identification of violated norms, graphical representations, or connectionism), using a running example taken from a corpus of car accident reports. Interestingly enough, the compared approaches focus on different aspects of the problem by either identifying all the potential causes, or selecting a smaller subset by taking advantages of contextually abnormal facts, or by modeling interventions to get rid of simple correlations. The paper concludes by a general discussion based on a battery of criteria (several of them being proper to AI approaches to causality).

Handling uncertainty and defeasibility in a possibilistic logic setting

Default rules express concise pieces of knowledge having implicit exceptions, which is appropriate for reasoning under incomplete information. Specific rules that explicitly refer to exceptions of more general default rules can then be handled in a non-monotonic setting. However, there is no assessment of the certainty with which the conclusion of a default rule holds when it applies. We propose a formalism in which uncertain default rules can be expressed, but still preserving the distinction between the defeasibility and uncertainty semantics by means of a two steps processing. Possibility theory is used for representing both uncertainty and defeasibility. The approach is illustrated in persistence modeling problems.

Update Postulates without Inertia

Starting from Katsuno and Mendelzon postulates, a new set of postulates which is minimal and complete has been defined, these postulates characterize an update operator. The update operator class which is obtained is larger than Katsuno and Mendelzons one, since it includes updating operators which are not inert, and which allow for the existence of unreachable states. The main property is that every update operator satisfying this new set of postulates admits an underlying ranked transition graph. The rank-ordering of its transitions has not to be faithful as in Katsuno and Mendelzon system.

Reasoning About Unpredicted Change and Explicit Time

Reasoning about unpredicted change consists in explaining observations by events; we propose here an approach for explaining time-stamped observations by surprises, which are simple events consisting in the change of the truth value of a fluent. A framework for dealing with surprises is defined. Minimal sets of surprises are provided together with time intervals where each surprise has occurred, and they are characterized from a model-based diagnosis point of view. Then, a probabilistic approach of surprise minimisation is proposed.

Formalizing Cognitive Acceptance of Arguments: Durum Wheat Selection Interdisciplinary Study

In this paper we present an interdisciplinary approach that concerns the problem of argument acceptance in an agronomy setting. We propose a computational cognitive model for argument acceptance based on the dual model system in cognitive psychology. We apply it in an agronomy setting within a French national project on durum wheat.