Showing papers by "Henri Prade published in 2005"

Interval-valued Fuzzy Sets, Possibility Theory and Imprecise Probability

Abstract: Interval-valued fuzzy sets were proposed thirty years ago as a natural extension of fuzzy sets. Many variants of these mathematical objects exist, under various names. One popular variant proposed by Atanassov starts by the specification of membership and non-membership functions. This paper focuses on interpretations of such extensions of fuzzy sets, whereby the two membership functions that define them can be justified in the scope of some information representation paradigm. It particularly focuses on a recent proposal by Neumaier, who proposes to use intervalvalued fuzzy sets under the name “clouds”, as an e! cient method to represent a family of probabilities. We show the connection between clouds, interval-valued fuzzy sets and possibility theory.

An argumentation-based approach to multiple criteria decision

Abstract: The paper presents a first tentative work that investigates the interest and the questions raised by the introduction of argumentation capabilities in multiple criteria decision-making. Emphasizing the positive and the negative aspects of possible choices, by means of arguments in favor or against them is valuable to the user of a decision-support system. In agreement with the symbolic character of arguments, the proposed approach remains qualitative in nature and uses a bipolar scale for the assessment of criteria. The paper formalises a multicriteria decision problem within a logical argumentation system. An illustrative example is provided. Various decision principles are considered, whose psychological validity is assessed by an experimental study.

On the representation, measurement, and discovery of fuzzy associations

Abstract: The use of fuzzy sets to describe associations between data extends the types of relationships that may be represented, facilitates the interpretation of rules in linguistic terms, and avoids unnatural boundaries in the partitioning of the attribute domains. In addition, the partial membership values provide a method for incorporating the distribution of the data into the assessment of a rule. This paper investigates techniques to identify and evaluate associations in a relational database that are expressible by fuzzy if-then rules. Extensions of the classical confidence measure based on the /spl alpha/-cut decompositions of the fuzzy sets are proposed to incorporate the distribution of the data into the assessment of a relationship and identify robustness in an association. A rule learning strategy that discovers both the presence and the type of an association is presented.

Towards a formal framework for the search of a consensus between autonomous agents

Abstract: This paper aims at proposing a general formal framework for dialogue between autonomous agents which are looking for a common agreement about a collective choice. The proposed setting has three main components: the agents, their reasoning capabilities, and a protocol. The agents are supposed to maintain beliefs about the environment and the other agents, together with their own goals. The beliefs are more or less certain and the goals may not have equal priority. These agents are supposed to be able to make decisions, to revise their beliefs and to support their points of view by arguments. A general protocol is also proposed. It governs the high-level behaviour of interacting agents. Particularly, it specifies the legal moves in the dialogue. Properties of the framework are studied. This setting is illustrated on an example involving three agents discussing the place and date of their next meeting.

Formal handling of threats and rewards in a negotiation dialogue

Abstract: Argumentation plays a key role in finding a compromise during a negotiation dialogue. It may lead an agent to change its goals/preferences and force it to respond in a particular way. Two types of arguments are mainly used for that purpose: threats and rewards. For example, if an agent receives a threat, this agent may accept the offer even if it is not fully "acceptable" for it (because otherwise really important goals would be threatened).The contribution of this paper is twofold. On the one hand, a logical setting that handles these two types of arguments is provided. More precisely, logical definitions of threats and rewards are proposed together with their weighting systems. These definitions take into account that negotiation dialogues involve not only agents' beliefs (of various strengths), but also their goals (having maybe different priorities), as well as the beliefs about the goals of other agents.On the other hand, a "simple" protocol for handling such arguments in a negotiation dialogue is given. This protocol shows when such arguments can be presented, how they are handled, and how they lead agents to change their goals and behaviors.

Encoding formulas with partially constrained weights in a possibilistic-like many-sorted propositional logic

Abstract: Possibilistic logic offers a convenient tool for handling uncertain or prioritized formulas and coping with inconsistency Propositional logic formulas are thus associated with weights belonging to a linearly ordered scale However, especially in case of multiple source information, only partial knowledge may be available about the relative ordering between weights of formulas In order to cope with this problem, a two-sorted counterpart of possibilistic logic is introduced Pieces of information are encoded as clauses where special literals refer to the weights Constraints between weights translate into logical formulas of the corresponding sort and are gathered in a distinct auxiliary knowledge base An inference relation, which is sound and complete with respect to preferential model semantics, enables us to draw plausible conclusions from the two knowledge bases The inference process is characterized by using "forgetting variables" for handling the symbolic weights, and hence an inference process is obtained by means of a DNF compilation of the two knowledge bases

Handling threats, rewards, and explanatory arguments in a unified setting

Abstract: Current logic-based handling of arguments has mainly focused on explanation or justification-oriented purposes in presence of inconsistency. So only one type of argument has been considered, and several argumentation frameworks have then been proposed for generating and evaluating such arguments. However, recent works on argumentation-based negotiation have emphasized different other types of arguments such as threats, rewards, and appeals. The purpose of this article is to provide a logical setting that encompasses the classical argumentation-based framework and handles the new types of arguments. More precisely, we give the logical definitions of these arguments and their weighting systems. These definitions take into account that negotiation dialogues involve not only agents' beliefs (of various strengths), but also their goals (having maybe different priorities), as well as the beliefs on the goals of other agents. In other words, from the different beliefs and goals bases maintained by agents, all the possible threats, rewards, explanations, and appeals that are associated with them can be generated. It may also happen that an intended threat, or reward, is not perceived as such by the addressee and thus misses its target because the addresser misrepresents the addressee's goals. The proposed approach accounts for that phenomenon. Finally, we show how to evaluate conflicting arguments of different types. © 2005 Wiley Periodicals, Inc. Int J Int Syst 20: 1195–1218, 2005.

A fuzzy set approach to concept-based information retrieval

Abstract: In this paper an information retrieval approach is proposed based on the use of a fuzzy conceptual structure used both to index document and to express user queries. The conceptual structure is hierarchical and it encodes the knowledge of the topical domain of the considered documents. It is formally represented as a weighted tree. The evaluation of conjunctive queries is based on the comparison of minimal subtrees containing the two sets of nodes corresponding to the concepts expressed in the document and the query respectively. The comparison uses different multiplevalued degrees of inclusion, which are discussed. The proposed approach generalizes standard fuzzy information retrieval. Its evaluation is also presented.

Rank-Ordering documents according to their relevance in information retrieval using refinements of ordered-weighted aggregations

Abstract: Classical information retrieval methods often lose valuable information when aggregating weights, which may diminish the discriminating power between documents. To cope with this problem, the paper presents an approach for ranking documents in IR, based on a vector-based ordering technique already considered in fuzzy logic for multiple criteria analysis purpose. Moreover, the proposed approach uses a possibilistic framework for evaluating queries to a document collection, which distinguishes between descriptors that are certainly relevant and those which are possibly relevant only. The proposal is evaluated on a benchmark collection that allows us to compare the effectiveness of this approach with a classical one. The proposed method provides an improvement of the precision w.r.t Mercure IR system.

An information-based discussion of vagueness: six scenarios leading to vagueness

Abstract: The issue of understanding and modeling vagueness has been addressed by many authors, especially in the second half of the twentieth century. In this chapter, we try to provide an organized discussion of different categories of vagueness, pointing out circumstances, where they appear. Together, they lead to a trichotomy of the universe of discourse, which seems to be the common feature of the different forms of vagueness. Basic representational frameworks are proposed for each case. This chapter does not advocate a particular view but identifies the characteristic features of each situation.

Expressing preferences from generic rules and examples – a possibilistic approach without aggregation function

Abstract: This paper proposes an approach to representing preferences about multifactorial ratings. Instead of defining a scale of values and aggregation operations, we propose to express rationality conditions and other generic properties, as well as preferences between specific instances, by means of constraints restricting a complete pre-ordering among tuples of values. The derivation of a single complete pre-order is based on possibility theory, using the minimal specificity principle. Some hints for revising a given preference ordering when new constraints are required, are given. This approach looks powerful enough to capture many aggregation modes, even some violating co-monotonic independence.

A bipolar possibilistic representation of knowledge and preferences and its applications

Abstract: When representing knowledge, it may be fruitful to distinguish between negative and positive information in the following sense. There are pieces of information ruling out what is known as impossible on the one hand, and pieces of evidence pointing out things that are guaranteed to be possible. But what is not impossible is not necessarily guaranteed to be possible. This applies as well to the modelling of the preferences of an agent when some potential choices are rejected since they are rather unacceptable, while others are indeed really satisfactory if they are available, leaving room for alternatives to which the agent is indifferent. The combination of negative information is basically conjunctive (as done classically in logic), while it is disjunctive in the case of positive information, which is cumulative by nature. This second type of information has been largely neglected by the logical tradition. Both types may be pervaded with uncertainty when modelling knowledge, or may be a matter of degree when handling preferences. The presentation will first describe how the two types of information can be accommodated in the framework of possibility theory. The existence of the two types of information can shed new light on the revision of a knowledge / preference base when receiving new information. It is also highly relevant when reasoning with (fuzzy) if-then rules, or for improving the expressivity of flexible queries.

Handling threats, rewards, and explanatory arguments in a unified setting: Research Articles

Abstract: Current logic-based handling of arguments has mainly focused on explanation or justification-oriented purposes in presence of inconsistency So only one type of argument has been considered, and several argumentation frameworks have then been proposed for generating and evaluating such arguments However, recent works on argumentation-based negotiation have emphasized different other types of arguments such as threats, rewards, and appeals The purpose of this article is to provide a logical setting that encompasses the classical argumentation-based framework and handles the new types of arguments More precisely, we give the logical definitions of these arguments and their weighting systems These definitions take into account that negotiation dialogues involve not only agents' beliefs (of various strengths), but also their goals (having maybe different priorities), as well as the beliefs on the goals of other agents In other words, from the different beliefs and goals bases maintained by agents, all the possible threats, rewards, explanations, and appeals that are associated with them can be generated It may also happen that an intended threat, or reward, is not perceived as such by the addressee and thus misses its target because the addresser misrepresents the addressee's goals The proposed approach accounts for that phenomenon Finally, we show how to evaluate conflicting arguments of different types © 2005 Wiley Periodicals, Inc Int J Int Syst 20: 1195–1218, 2005

Improving document ranking in information retrieval using ordered weighted aggregation and leximin refinement

Abstract: Classical information retrieval (IR) methods often lose valuable information when aggregating weights, which may diminish the discriminating power between documents To cope with this problem, the paper presents an approach for ranking documents in IR, based on a vector-based ordering technique already considered in fuzzy logic for multiple criteria analysis purpose Moreover, the proposed approach uses a possibilistic framework for encoding the retrieval status values The approach, applied to a benchmark collection, has been shown to improve IR precision wrt classical approaches

Coping with exceptions in multiclass ILP problems using possibilistic logic

Abstract: The handling of exceptions in multiclass problems is a tricky issue in inductive logic programming (ILP) In this paper we propose a new formalization of the ILP problem which accounts for default reasoning, and is encoded with first-order possibilistic logic We show that this formalization allows us to handle rules with exceptions, and to prevent an example to be classified in more than one class The possibilistic logic view of ILP problem, can be easily handled at the algorithmic level as an optimization problem

Flexible Querying with Argued Answers

Abstract: The paper presents a preliminary work that investigates the interest and the questions raised by the introduction of argumentation capabilities in a flexible querying system to a database. Indeed, emphasizing the positive and the negative aspects of retrieved items, or explaining why no answers are found may be of a value for the user

Possibilistic inductive logic programming

Abstract: Learning rules with exceptions may be of interest, especially if the exceptions are not important in some sense. Standard Inductive Logic Programming (ILP) algorithms and classical first order logic are not well-suited for managing rules with exceptions. Indeed, a hypothesis that is induced accumulates all the exceptions of the rules contained in it. Moreover, with multiple-class problems, classifying an example in two different classes (even if one is the right one) is not correct, so a rule that contains some exceptions may prevent another rule which has no exception from being useful. This paper proposes a new possibilistic logic framework for weighted ILP. It induces rules which are progressively more and more accurate, and allows us to manage exceptions by controlling their accumulation. In this setting, we first propose an algorithm for learning rules when the background knowledge and the examples are stratified into layers having different levels of priority or certainty. This allows the induction of general but uncertain rules together with more specific and less uncertain rules. A second algorithm is presented, which does not require an initial weighted database, but still learn a default set of rules in the possibilistic setting.

Fuzzy Inductive Logic Programming: Learning Fuzzy Rules with their Implication

Abstract: Inductive logic programming (ILP) is a generic tool aiming at learning rules from relational databases. Introducing fuzzy sets arid fuzzy implication connectives in this framework allows us to increase the expressive power of the induced rules while keeping the readability of the rules. Moreover, fuzzy sets facilitate the handling of numerical attributes by avoiding crisp and arbitrary transitions between classes. In this paper, the meaning of a fuzzy rule is encoded by its implication operator, which is to be determined in the learning process. An algorithm is proposed for inducing first order rules having fuzzy predicates, together with the most appropriate implication operator. The benefits of introducing fuzzy logic in ILP and the validation process of what has been learnt are discussed and illustrated on a benchmark

Formal handling of threats and rewards in a negotiation dialogue

Abstract: Argumentation plays a key role in finding a compromise during a negotiation dialogue. It may lead an agent to change its goals/ preferences and force it to respond in a particular way. Two types of arguments are mainly used for that purpose: threats and rewards. For example, if an agent receives a threat, this agent may accept the offer even if it is not fully “acceptable” for it (because otherwise really important goals would be threatened). The contribution of this paper is twofold. On the one hand, a logical setting that handles these two types of arguments is provided. More precisely, logical definitions of threats and rewards are proposed together with their weighting systems. These definitions take into account that negotiation dialogues involve not only agents’ beliefs (of various strengths), but also their goals (having maybe different priorities), as well as the beliefs about the goals of other agents. On the other hand, a “simple” protocol for handling such arguments in a negotiation dialogue is given. This protocol shows when such arguments can be presented, how they are handled, and how they lead agents to change their goals and behaviors.

Acceptance, conditionals, and belief revision

Abstract: This paper bridges the gap between comparative belief structures, such as those induced by probability measures, and logical representations of accepted beliefs. We add, to natural properties of comparative belief relations, some conditions that ensure that accepted beliefs form a deductively closed set. It is shown that the beliefs accepted by an agent in all contexts can always be described by a family of conditionals. These results are closely connected to the nonmonotonic 'preferential' inference system of Kraus, Lehmann and Magidor and the works of Friedman and Halpern on their so-called plausibility functions. Acceptance relations are also another way of approaching the theory of belief change after the works of Gardenfors and colleagues.

Qualitative evaluation of decisions in an argumentative manner - A general discussion in a unified setting

Abstract: The paper intends to provide a unified discussion of different types of decision processes (decision under uncertainty, multiple-criteria decision, case-based decision, rule-based decision) from an argumentation point of view. This means here that in the evaluation of a decision, we carefully distinguish what positively contributes to its evaluation and what negatively contributes to it. The discussion favors a qualitative evaluation setting, which uses a bivariate scale for assessing the values of consequences. The nature of arguments changes with the type of decision, as well as how their strength can be assessed. However, a preliminary view of how balancing pros and cons for ranking decisions is proposed. Some illustrative examples are provided.

Towards a formal framework for the search of a consensus between autonomous agents

Abstract: This paper aims at proposing a general formal framework for dialogue between autonomous agents which are looking for a common agreement about a collective choice. The proposed setting has three main components: the agents, their reasoning capabilities, and a protocol. The agents are supposed to maintain beliefs about the environment and the other agents, together with their own goals. The beliefs are more or less certain and the goals may not have equal priority. These agents are supposed to be able to make decisions, to revise their beliefs and to support their points of view by arguments. A general protocol is also proposed. It governs the high-level behaviour of interacting agents. Particularly, it specifies the legal moves in the dialogue. Properties of the framework are studied. This setting is illustrated on an example involving three agents discussing the place and date of their next meeting.