In order to avoid ambiguity and be efficient, the context in which a query is made can help to better target the relevant pieces of information from the knowledge base to be processed by the inference system. In this paper, we are interested in the notion of dynamical compartmentalization where the knowledge base that will be used for reasoning is dynamically extracted from the original base. Compartmentalization is a selection of a sub-base which is done according to a function, called refiner, and depending on this function some properties are satisfied. We introduce a particular syntactic refiner that uses a similarity symbol-based distance between a context (a multiset of variable symbols) and a formula of a knowledge base. We prove that the inference operator based on this refiner, called contextual inference, satisfies a series of desirable axioms

History eXplanation based on Predicates (HXP), studies the behavior of a Reinforcement Learning (RL) agent in a sequence of agent’s interactions with the environment (a history), through the prism of an arbitrary predicate [21]. To this end, an action importance score is computed for each action in the history. The explanation consists in displaying the most important actions to the user. As the calculation of an action’s importance is #W[1]-hard, it is necessary for long histories to approximate the scores, at the expense of their quality. We therefore propose a new HXP method, called Backward-HXP, to provide explanations for these histories without having to approximate scores. Experiments show the ability of B-HXP to summarise long histories.

International audience

We provide a unified framework in which the three emotions at the heart of narrative tension (curiosity, suspense and surprise) are formalized. This framework is built on non-monotonic reasoning which allows us to compactly represent the default behavior of the world and to simulate the affective evolution of an agent receiving a story. After formalizing the notions of awareness, curiosity, surprise and suspense, we explore the properties induced by our definitions and study the computational complexity of detecting them. We finally propose means to evaluate these emotions’ intensity for a given agent listening to a story.

best student paper award

A bipolar structure called BLF expresses knowledge about decisions in terms of decision principles that are ranked and polarized according to the utility of the consequences of these decisions. A BLF allows us to compare decisions under incomplete knowledge. For a given decision, the BLF returns a vector of utility/dis-utility in terms of achievement of positive/negative goals. Decisions are compared thanks to these vectors. In this paper we focus on the link between the uncertain knowledge aggregation made by the BLF and classical aggregation functions used in decision under uncertainty and multi-criteria approaches. The main benefit of a BLF is that thanks to the bipolar scale, positive and negative goals can be dealt with independently under their own point of view (each of them being either pessimistic or optimistic).

The paper deals with a topic little studied in artificial intelligence: the understanding of humor. In this preliminary study, we try to identify the basic mechanism at work in quips and narrative jokes. It seems that in many cases a belief revision process is operating, leading to an unexpected conclusion, through the punchline of the jest. We propose a formal modeling of jokes based on belief revision. Namely the punchline, which triggers a revision, is both surprising and explains perfectly what was reported in the beginning of the joke. This also suggests a way of ranking jokes in terms of surprise and strength of explanation, using possibilistic logic.

Our research concerns the problem of explainable decision in a context of incomplete knowledge. We define a framework called Bipolar Layered Framework with Support and Weights (BLFSW) that represents the set of argument graphs that can be used in the domain, enabling us to compute what results can be obtained in the different decision situations. This framework also contains information about the utilities/disutilities of these tangible results. This paper extends Bipolar Layered Frameworks defined in [1] by enabling the expression of supports for decision principles and by giving the user the possibility to fix the strength of inhibitors and supports with weights. This increased expressiveness of the framework is important both for refining the evaluation of alternatives and to improve the compactness of the representation. The main result of this paper is to provide an automatic way to explain a possibilistic decision setting in terms of a BLFSW which makes explicit the principles that govern the decision.

This paper presents a set of principles that an intuitive and efficient visual representation language should satisfy. Then after a presentation of the visual typed language MOT, we show that MOT may be criticized which leads us to introduce an improvement of MOT called VTL. VTL is a Visual Typed Language satisfying most of the principles that we introduced.

In this paper we approach both procedural and substantive irrationality of artificial agent cognitive systems and consider that when it is not possible for an agent to make a logical inference (too expensive cognitive effort or not enough knowledge) she might replace certain parts of the logical reasoning with mere associations.

The purpose of this paper is to describe some results of the LELIE project, that are a contribution of Artificial Intelligence to a special domain: the analysis of the risks due to poorly written technical documents. This is a multidisciplinary contribution since it combines natural language processing with logical satisfiability checking. This paper explains how satisfiability checking can be used for detecting inconsistencies, redundancy and incompleteness in procedural texts and describes the part of the implemented tool that produces the logical translation of technical texts and realizes the checkings.

Several systems were developed for supporting public persuasion dialogs where two agents with conflicting opinions try to convince an audience. For computing the outcomes of dialogs, these systems use (abstract or structured) argumentation systems that were initially developed for nonmonotonic reasoning. Despite the increasing number of such systems, there are almost no work on high level properties they should satisfy. This paper is a first attempt for defining postulates that guide the well-definition of dialog systems and that allow their comparison. We propose six basic postulates (including e.g. the finiteness of generated dialogs). We then show that this set of postulates is incompatible with those proposed for argumentation systems devoted for nonmonotonic reasoning. This incompatibility confirms the differences between persuading and reasoning. It also suggests that reasoning systems are not suitable for computing the outcomes of dialogs.

This paper defines a new framework for dynamics in argumentation. In this framework, an agent can change an argumentation system (the target system) in order to achieve some desired goal. Changes consist in addition/removal of arguments or attacks between arguments and are constrained by theagent's knowledge encoded by another argumentation system. We present a software that computes the possible change operations for a given agent on a given target argumentation system in order to achieve some given goal.

In this paper, we address the problem of change in an abstract argumentation system. We focus on a particular change: the addition of a new argument which interacts with previous arguments. We study the impact of such an addition on the outcome of the argumentation system, more particularly on the set of its extensions. Several properties for this change operation are defined by comparing the new set of extensions to the initial one, these properties are called structural when the comparisons are based on set-cardinality or set-inclusion relations. Several other properties are proposed where comparisons are based on the status of some particular arguments: the accepted arguments; these properties refer to the evolution of this status during the change, e.g., Monotony and Priority to Recency. All these properties may be more or less desirable according to specific applications. They are studied under two particular semantics: the grounded and preferred semantics.

In complex multiagent systems, the agents may be heterogeneous and possibly designed by different programmers. Thus, the importance of defining a standard framework for agent communication languages (ACL) with a clear semantics has been widely recognized. The semantics should be verifiable, clear, and practical. Most classical proposals (for instance, mentalistic semantics) fail to meet these objectives. This paper proposes a logic-based semantics, which is social in nature. The basic idea is to associate with each speech act a clear meaning in terms of a commitment induced by that speech act, and a penalty to be paid in case that commitment is violated. A violation criterion based on the existence of arguments is then defined per speech act. We show that the proposed semantics satisfies some key properties that ensure that the approach is well founded. The logical setting makes the semantics verifiable. Moreover, it is shown that the new semantics is practical because it captures the dynamic of dialogues and shows clearly how isolated speech acts can be connected for building dialogues. © 2008 Wiley Periodicals, Inc.