Search

A large number of evaluation methods, called semantics, have been proposed in the literature for assessing strength of arguments. This paper investigates their equivalence. It argues that for being equivalent, two semantics should have compatible evaluations of both individual arguments and pairs of arguments. The first requirement ensures that the two semantics judge an argument in the same way, while the second states that they provide the same ranking of arguments. We show that the two requirements are completely independent. The paper introduces three novel relations between semantics based on their rankings of arguments: weak equivalence, strong equivalence and refinement. They state respectively that two semantics do not disagree on their strict rankings; the rankings of the semantics coincide; one semantics agrees with the strict comparisons of the second and it may break some of its ties. We investigate the properties of the three relations and their links with existing principles of semantics, and study the nature of relations between most of the existing semantics. The results show that the main extensions semantics are pairwise weakly equivalent. The gradual semantics we considered are pairwise incompatible, however some pairs are strongly equivalent in case of flat graphs including Max-based (Mbs) and Euler-based (Ebs), for which we provide full characterizations in terms respectively of Fibonacci numbers and the numbers of an exponential series. Furthermore, we show that both semantics (Mbs, EMbs) refine the grounded semantics, and are weakly equivalent with the other extension semantics. We show also that in case of flat graphs, the two gradual semantics Trust-based and Iterative Schema characterize the grounded semantics, making thus bridges between gradual semantics and extension semantics. Finally, the other gradual semantics are incompatible with extension semantics.

International audience; This chapter is devoted to logical models for reasoning from contradictory information. It deals with methods, such as argumentation, that refrain from giving up any piece of information (by contrast with revision, as discussed in chapter “Main Issues in Belief Revision, Belief Merging and Information Fusion” of this volume). The baseline is to get the best, resorting to various possibilities, from the available information in order to reason in the most sensible way despite contradictions.

This chapter presents several techniques allowing agents to come up with an agreement. We start by discussing negotiation among two agents: after having recalled the axiomatic approach of Nash, we present a standard protocol, and point to recent advances in the field. We then discuss issues raised in the multilateral case. Finally, we conclude the chapter by describing an example of persuasion-based negotiation, where agents can put forward justifying reasons through the negotiation, so as to possibly modify preferences over offers or more generally, influence the negotiation process.

The paper tackled the issue of arguments evaluation in weighted bipolar argumen-tation graphs (i.e., graphs whose arguments have basic strengths, and may be both supported and attacked). We introduce principles that an evaluation method (or semantics) could satisfy. Such principles are very useful for understanding the foundations of semantics, judging them, and comparing semantics. We then analyze existing semantics on the basis of our principles, and finally propose a new semantics for the class of acyclic graphs. We show that it satisfies all the principles.

Rule-based argumentation systems are developed for reasoning about defeasible information. As a major feature, their logical language distinguishes between strict rules (encoding strict information) and defeasible rules (describing general behavior with exceptional cases). They build arguments by chaining such rules, define attacks between them, use a semantics for evaluating the arguments and finally identify the plausible conclusions that follow from the rules. Focusing on the family of inconsistency-based attack relations, this paper presents the first study of the outcomes of such systems under various acceptability semantics, namely naive, stable, semi-stable, preferred, grounded and ideal. It starts by extending the existing list of rationality postulates that any rule-based system should satisfy. Then, it defines the key notion of option of a theory (a theory being a set of facts, a set of strict rules and a set of defeasible rules). For each of the cited semantics, it characterizes the extensions of a rule-based system that satisfies all the postulates in terms of options of the theory under which the system is built. It also fully characterizes the set of plausible conclusions of the system. The results show that designing a rule-based argumentation system requires great care.

Rule-based argumentation systems are developed for reasoning about defeasible information. They take as input a theory made of a set of facts , a set of strict rules , which encode strict information, and a set of defeasible rules which describe general behavior with exceptional cases. They build arguments by chaining such rules, define attacks between them, use a semantics for evaluating the arguments, and finally identify the plausible conclusions that follow from the theory. Undercutting is one of the main attack relations of such systems. It consists of blocking the application of defeasible rules when their exceptional cases hold. In this paper, we consider this relation for capturing all the different conflicts in a theory. We present the first argumentation system that uses only undercutting, and show that it satisfies the rationality postulates proposed in the literature. Finally, we fully characterize both its extensions and its plausible conclusions under various acceptability semantics. Indeed, we show full correspondences between extensions and sub-theories of the theory under which the argumentation system is built.

International audience; Recently, the notion of similarity between arguments, namely those built using propositional logic, has been investigated and several similarity measures have been defined. This paper shows that those measures may lead to inaccurate results when arguments are not concise, i.e., their supports contain information that is useless for inferring their conclusions. For circumventing this limitation, we start by refining arguments for making them concise. Then, we propose two families of similarity measures that extend existing ones and that deal with concise arguments.

This is an abridged version of a paper titled “Evaluation of ar-guments on weighted bipolar graphs” which won a best-paper awardat ECSQARU-17 conference.; International audience; The paper studies how arguments can be eval-uated inweighted bipolar argumentation graphs(i.e., graphs whose arguments have basic weightsand may be supported and attacked). It introducesprinciplesthat an evaluation method (orsemantics)would satisfy, analyzes existing semantics with re-spect to them, and finally proposes a new semanticsfor the class of non-maximal acyclic graph

International audience; The paper studies semantics that evaluate arguments in argumentation graphs, where each argument has a basic strength, and may be attacked by other arguments. It starts by defining a set of principles , each of which is a property that a semantics could satisfy. It provides the first formal analysis and comparison of existing semantics. Finally, it defines three novel semantics that satisfy more principles than existing ones.

International audience; This paper investigates when two abstract logic-based argumentation systems are equivalent. It defines various equivalence criteria, investigates the links between them, and identifies cases where two systems are equivalent with respect to each of the proposed criteria. In particular, it shows that under some reasonable conditions on the logic underlying an argumentation system, the latter has an equivalent finite subsystem, called core. This core constitutes a threshold under which arguments of the system have not yet attained their final status and consequently adding a new argument may result in status change. From that threshold, the statuses of all arguments become stable.

International audience; Dung’s (1995) argumentation framework takes as input two abstract entities: a set of arguments and a binary relation encoding attacks between these arguments. It returns acceptable sets of arguments, called extensions, w.r.t. a given semantics. While the abstract nature of this setting is seen as a great advantage, it induces a big gap with the application that it is used to. This raises some questions about the compatibility of the setting with a logical formalism (i.e., whether it is possible to instantiate it properly from a logical knowledge base), and about the significance of the various semantics in the application context. In this paper we tackle the above questions. We first propose to fill in the previous gap by extending Dung’s (1995) framework. The idea is to consider all the ingredients involved in an argumentation process. We start with the notion of an abstract monotonic logic which consists of a language (defining the formulas) and a consequence operator. We show how to build, in a systematic way, arguments from a knowledge base formalised in such a logic. We then recall some basic postulates that any instantiation should satisfy. We study how to choose an attack relation so that the instantiation satisfies the postulates. We show that symmetric attack relations are generally not suitable. However, we identify at least one ‘appropriate’ attack relation. Next, we investigate under stable, semi-stable, preferred, grounded and ideal semantics the outputs of logic-based instantiations that satisfy the postulates. For each semantics, we delimit the number of extensions an argumentation system may have, characterise the extensions in terms of subsets of the knowledge base, and finally characterise the set of conclusions that are drawn from the knowledge base. The study reveals that stable, semi-stable and preferred semantics either lead to counter-intuitive results or provide no added value w.r.t. naive semantics. Besides, naive semantics either leads to arbitrary results or generalises the coherence-based approach initially developed by Rescher and Manor (1970). Ideal and grounded semantics either coincide and generalise the free consequence relation developed by Benferhat, Dubois, and Prade (1997), or return arbitrary results. Consequently, Dung’s (1995) framework seems problematic when applied over deductive logical formalisms.

Argumentation is a promising approach for defeasible reasoning. It consists of justifying each plausible conclusion by arguments. Since the available information may be inconsistent, a conclusion and its negation may both be justified. The arguments are thus said to be conflicting. The main issue is how to evaluate the arguments. Several semantics were proposed for that purpose. The most important ones are: stable, preferred, complete, grounded and admissible. A semantics is a set of criteria that should be satisfied by a set of arguments, called extension, in order to be acceptable. Different decision problems related to these semantics were defined (like whether an argumentation framework has a stable extension). It was also shown that most of these problems are intractable. Consequently, developing algorithms for these problems is not trivial and thus the implementation of argumentation systems not obvious. Recently, some solutions to this problem were found. The idea is to use a reduction method where a given problem is translated in another one like SAT or ASP. This paper follows this line of research. It studies how to encode the problem of computing the extensions of an argumentation framework (under each of the previous semantics) as a constraint satisfaction problem (CSP). Such encoding is of great importance since it makes it possible to use the very efficient solvers (developed by the CSP community) for computing the extensions. Our encodings take advantage of existing reductions to SAT problems in the case of Dung's abstract framework. Among the various ways of translating a SAT problem into a CSP one, we propose the most appropriate one in the argumentation context. We also provide encodings in case two other families of argumentation frameworks: the constrained version of Dung's abstract framework and preference-based argumentation framework.

This article proposes an abstract framework for argumentation-based negotiation in which the impact of exchanging arguments on agents' theories is formally described, the different types of solutions in negotiation are investigated, and the added value of argumentation in negotiation dialogues is analysed. We study when, how and to which extent an exchange of arguments can be beneficial in negotiation. The results show that argumentation can improve the quality of an outcome but never decrease it.

Dung's argumentation framework consists of a set of arguments and an attack relation among them. Arguments are evaluated and acceptable sets of them, called extensions, are computed using a given semantics. Each extension represents a coherent position. In the literature, several proposals have extended this framework in order to take into account the strength of arguments. The basic idea is to ignore an attack if the attacked argument is stronger than (or preferred to) its attacker. Semantics are then applied using only the remaining attacks. In this paper, we show that those proposals behave correctly when the attack relation is symmetric. However, when it is asymmetric, conflicting extensions may be computed leading to unintended conclusions. We propose an approach that guarantees conflict-free extensions. This approach presents two novelties: the first one is that it takes into account preferences at the semantics level rather than the attack level. The idea is to extend existing semantics with preferences. In case preferences are not available or do not conflict with the attacks, the extensions of the new semantics coincide with those of the basic ones. The second novelty of our approach is that a semantics is defined as a dominance relation on the powerset of the set of arguments. The extensions (under a semantics) are the maximal elements of the dominance relation. Such an approach makes it possible not only to compute the extensions of a framework but also to compare its non-extensions. We start by proposing three dominance relations that generalize respectively stable, preferred and grounded semantics with preferences. Then, we focus on stable semantics and provide full characterizations of its dominance relations and those of its generalized versions. Complexity results are provided. Finally, we show that an instance of the proposed framework retrieves the preferred sub-theories which were proposed in the context of handling inconsistency in weighted knowledge bases.

Decision making is usually based on the comparative evaluation of different options by means of a decision criterion. Recently, the qualitative pessimistic criterion was articulated in terms of a four-step argumentation process: (i) to build arguments in favour/against each option, (ii) to compare and evaluate those arguments, (iii) to assign a status for each option, and (iv) to rank order the options on the basis of their status. Thus, the argumentative counter-part of the pessimistic criterion provides not only the ‘best’ option to the user but also the reasons justifying this recommendation. The aim of this article is to study the dynamics of this argumentation model. The idea is to study how the ordering on options changes in light of a new argument. For this purpose, we study under which conditions an option may change its status, and under which conditions the new argument has no impact on the status of options, and consequently, on the ordering. This amounts to study how the acceptability of arguments evolves when the decision system is extended by new arguments. In the article, we focus on two acceptability semantics the skeptical grounded semantics and the credulous preferred semantics.

International audience; Argumentation is a promising approach used by autonomous agents for reasoning about inconsistent/incomplete/uncertain knowledge, based on the construction and the comparison of arguments. In this paper, we apply this approach to the classification problem, whose purpose is to construct from a set of training examples a model that assigns a class to any new example. We propose a formal argumentation-based model that constructs arguments in favor of each possible classification of an example, evaluates them, and determines among the conflicting arguments the acceptable ones. Finally, a “valid” classification of the example is suggested. Thus, not only the class of the example is given, but also the reasons behind that classification are provided to the user as well in a form that is easy to grasp. We show that such an argumentation-based approach for classification offers other advantages, like for instance classifying examples even when the set of training examples is inconsistent, and considering more general preference relations between hypotheses. In the particular case of concept learning, the results of version space theory developed by Mitchell are retrieved in an elegant way in our argumentation framework. Finally, we show that the model satisfies the rationality postulates identified in argumentation literature. This ensures that the model delivers sound results.

An argument is seen as reason in favour of a claim. It is made of three parts: a set of premises representing the reason, a conclusion representing the supported claim, and a connection showing how the premises lead to the conclusion. Arguments are frequently exchanged by human agents in natural language (spoken or written) in discussion, debate, negotiation, persuasion, etc. They may be very different in that their three components may have various forms. In this paper, we propose a language for representing such arguments. We show that it is general enough to capture the various forms of arguments encountered in natural language, and that it is possible to represent attack and support relations between arguments as formulas of the same language.

International audience; Rule-based argumentation systems are developed for reasoning about defeasible information. They take as input a theory made of a set of strict rules, which encode strict information, and a set of defeasible rules which describe general behaviour with exceptional cases. They build arguments by chaining such rules, define attacks between them, use a semantics for evaluating the arguments, and finally identify the plausible conclusions that follow from the rules. One of the main attack relations of such systems is the so-called undercutting which blocks the application of defeasible rules in some contexts. In this paper, we show that this relation is powerful enough to capture alone all the different conflicts in a theory. We present the first argumentation system that uses only undercutting and fully characterize both its extensions and its plausible conclusions under various acceptability semantics.

Bulletin AFIA

International audience; Logic-based argumentation systems are developed for reasoning with inconsistent information. Starting from a knowledge base encoded in a logical language, they define arguments and attacks between them using the consequence operator associated with the language. Finally, a semantics is used for evaluating the arguments. In this paper, we focus on systems that are based on deductive logics and that use Dung's semantics. We investigate rationality postulates that such systems should satisfy. We define five intuitive postulates: consistency and closure under the consequence operator of the underlying logic of the set of conclusions of arguments of each extension, closure under sub-arguments and exhaustiveness of the extensions, and a free precedence postulate ensuring that the free formulas of the knowledge base (i.e., the ones that are not involved in inconsistency) are conclusions of arguments in every extension. We study the links between the postulates and explore conditions under which they are guaranteed or violated.

This paper studies argumentation-based dialogues between agents. It defines a set of locutions by which agents can trade arguments, a set of agent attitudes which relate what arguments an agent can build and what locutions it can make, and a set of protocols by which dialogues can be carried out. The paper then considers some properties of dialogues under the protocols, in particular termination, dialogue outcomes, and complexity, and shows how these relate to the agent attitudes.

We propose a dialogue game protocol for purchase negotiation dialogues which identifies appropriate speech acts, defines constraints on their utterances, and specifies the different sub-tasks agents need to perform in order to engage in dialogues according to this protocol. Our formalism combines a dialogue game similar to those in the philosophy of argumentation with a model of rational consumer purchase decision behaviour adopted from marketing theory. In addition to the dialogue game protocol, we present a portfolio of decision mechanisms for the participating agents engaged in the dialogue and use these to provide our formalism with an operational semantics. We show that these decision mechanisms are sufficient to generate automated purchase decision dialogues between autonomous software agents interacting according to our proposed dialogue game protocol.

Argumentation is a promising approach to handle inconsistent knowledge bases, based on the justification of plausible conclusions by arguments. Because of inconsistency, however, arguments may be defeated by counterarguments (or defeaters). The problem is thus to select the most acceptable arguments. In this paper we investigate preference-based acceptability. The basic idea is to accept undefeated arguments and also arguments that are preferred to their defeaters. We say that these arguments defend themselves against their defeaters. We define argumentation frameworks based on that preference-based acceptability. Finally, we study associated inference relations for reasoning with inconsistent knowledge bases.

Argumentation is a reasoning model based on the construction of arguments and counter-arguments (or defeaters) followed by the selection of the most acceptable of them. In this paper, we refine the argumentation framework proposed by Dung by taking into account preference relations between arguments in order to integrate two complementary points of view on the concept of acceptability: acceptability based on the existence of direct counter-arguments and acceptability based on the existence of defenders. An argument is thus acceptable if it is preferred to its direct defeaters or if it is defended against its defeaters. This also refines previous works by Prakken and Sartor, by associating with each argument a notion of strength, while these authors embed preferences in the definition of the defeat relation. We propose a revised proof theory in terms of AND/OR trees, verifying if a given argument is acceptable, which better reflects the dialectical form of argumentation.