LAF (London Argumentation Forum)

Start Date: 2011-03-25

End Date: 2011-03-25

Location: London

Country: UK

Description:

LAF, the London Argumentation Forum, is organised as part of the activities of the Agreement Technology COST action, working group WG4. The forum is open to anyone working in argumentation and interested in the activities of WG4.

LAF is intended to provide an opportunity for PhD students and early researchers to present, discuss and get feedback on their recent work on argumentation. Participants will be invited to give brief presentations, followed by question sessions intended to stimulate discussion during the course of the day, and continuing into the evening, during which a social event will be organised.
The forum will take place on Friday 25 March at Imperial College London, Room 311 in Huxley Building.

We have a limited number of grants available for participants, especially if from outside the UK and PhD students/early researchers.
If you are interested in participating, please send an email to Francesca Toni (ft@imperial.ac.uk) by 20 February, indicating:

  1. Your name and affiliation
  2. Your position (PhD student/early researcher/non-early researcher)
  3. If you would like to give a presentation: title and short abstract of your proposed presentation
  4. If you would like to participate without giving a presentation: a brief description of your research interests/activities in argumentation
  5. Whether you will require funding and an estimate of this funding.

Preliminary programme

Stella Heras
Talk: A Case-Based Argumentation Framework for Agent Societies

Abstract: The capability of reaching agreements is a necessary feature that  large computer systems where agents interoperate must include. In  these systems, agents represent self-motivated entities that have a  social context, including dependency relations among them, and  different preferences and beliefs. Without agreement there is no cooperation and thus, complex tasks which require the interaction of agents with different points of view cannot be performed. In this work, we propose a case-based argumentation framework for the design and implementation of MAS where agents reach agreements by arguing.  The knowledge resources and the reasoning process that agents use to manage their positions and arguments are presented. In addition, the  framework allows agents to improve their argumentation skills from experience.

 

Wolfgang Dvorak
Talk: On the Intertranslatability of Argumentation Semantics

Abstract: Translations between different nonmonotonic formalisms always have been an important topic, in particular to understand the knowledge representation capabilities those formalisms offer. We provide such an investigation in terms of different semantics proposed for abstract argumentation frameworks. Although the properties of these different semantics are nowadays well understood, there are no explicit results about intertranslatability. We provide such translations wrt. different properties and also give some negative results that show that certain translations are not possible.

Cristian Gratie
Talk: Research Ideas for Extending Argument Labeling

Abstract: I will start by reviewing complete labelings and their relation to extension-based semantics (the work of Caminada). I will then talk about extending the number of labels: why would we do that? how would we use such labelings to get back to extensions? Then I will talk about real-valued labels and give some interesting results I have obtained about them (since this is part of a paper currently under review for ArgMAS, I am not sure whether I’ll give too many details or just mention the general idea). I’ll end the presentation by pointing out that it may be interesting to look for new rules for labeling so as to relate the labels with the justification state of arguments.

  

Federico Cerutti

 Talk: Argumentation-based Decision Support for Practical Reasoning

 

Abstract:  This talk describes a proposal of an argumentation-based approach to  modeling articulated decision support contexts. The proposed approach encompasses a variety of argument and attack schemes aimed at

representing basic knowledge and reasoning patterns for decision support. Some of the defined attack

schemes involve attacks directed towards other attacks, which are not allowed in traditional argumentation frameworks but turn out to be useful as a knowledge and reasoning modeling tool: in particular, we

demonstrate their use to support what-if reasoning capabilities, which are of primary importance in decision support. Formal backing to this approach is provided by the AFRA formalism, a recently proposed

extension of Dung’s argumentation framework. 

 

Xiuyi Fan

 Talk: Conflict resolution with argumentation dialogues

 

Abstract: Conflicts exist in multi-agent systems for a number of reasons: agents have different interests and desires; agents hold different beliefs; agents make different assumptions. To resolve conflicts, agents need to better convey information to each other and facilitate fair negotiations yielding jointly agreeable outcomes.  We present a two-agent, dialogical conflict resolution scheme developed with the Assumption-Based Argumentation (ABA) framework.

 

 Radu-Casian Mihailescu

 Talk: An organizational approach for efficient dynamic coalition formation in stochastic environments

 

Abstract: The research issue I essentially investigate regards grouping techniques formed amongst rational agents. From the argumentation point of view, the challenge is twofold. Primarily, the initiator agent must identify potential group members and persuade them to cooperate in achieving the goal concerned. Secondly, the group members need to deliberate about the appropriate plan that would bring about the initial goal. To this extent, we intend to integrate such argumentation schemes and design interaction protocols accordingly, for specific game settings, then analyze the notion of stability that the system at large is capable to achieve. Thus, finally, the question we pose is whether coherent dynamics of groups of agents can spontaneously emerge via multi-party deliberation dialogues and how these protocols may be manipulated in order to enforce certain desired coherent patterns of agents’ group activity.

 

 Bas van Gijzel

 Talk: Relating Carneades with Abstract Argumentation

 

Abstract: Carneades is a recently proposed formalism for structured argumentation with varying proof standards. While the formalism is very interesting, to date its relation with Dung’s seminal abstract approach to argumentation is unknown. In this talk the two formalisms are formally related by translating Carneades into ASPIC+, which is another recently proposed formalism for structured argumentation by Henry Prakken. Since ASPIC+ is defined to generate abstract argumentation frameworks, which are the input of Dung’s approach, a translation of Carneades into ASPIC+ provides a way to translate Carneades graphs into abstract argumentation frameworks. In particular, it will be proven that Carneades always induces a unique Dung extension, which is the same in all of Dung’s semantics.

 

Rolando Medellin Gasque

 Talk: Arguments over Plans

 

Abstract: Autonomous planning agents that share a common goal should be able to propose, justify and share information about plans. To reach an agreement on the best plan to follow agents need to apply strategies to persuade and negotiate with other agents to resolve inconsistencies about the world. We present an argumentation

scheme and associated critical questions to create and justify plan proposals where plans are combination of actions. An analysis of different action combination is presented and then linked with the argumentation scheme and the critical questions. We believe these elements are necessary to enable agents to engage in rational debate over plan proposals.

 

 

Martyn Lloyd-Kelly

 Talk: Practical reasoning argumentation and emotion

 

Abstract:  One important feature of practical reasoning concerns what other agents will do. This requires consideration of emotional aspects, as well as the agents’ interests. In this presentation I will discuss a framework under development for investigating the effects of emotional responses upon agents’ decision making. I will describe an implementation test-bed, using the TileWorld domain, which takes inspiration from game theory, in particular Axelrod’s tournament for investigating strategies in the Prisoner’s Dilemma. I will indicate how the experiments to be conducted will investigate agents’ emotional responses to the actions of others and how reasoning about this can be captured through the use of a practical reasoning argumentation scheme extended to incorporate the notion of emotion.

  

Alan Perotti

 Talk: Argumentative Agents Negotiating on Potential Attacks

 

Abstract: Real-life situations where agents want to gradually  reveal the information they own and discuss about the details after  agreeing about the general problems can be modeled using an extended  argumentation framework with potential attacks. In multiagent systems,  the agents negotiate about raising potential attacks or not, in order

to get the maximal outcome, expressed as the number of accepted arguments. Moreover, agents evaluate the benefits of the negotiation  proposals due to their utility functions based, for instance, on the  preferences about the arguments. The result of the negotiation process  consists in the formation of coalitions composed by those agents which  have found an agreement. The two proposed negotiation protocols have  been implemented and an evaluation, addressed by means of experimental  results, shows which combination of strategies and negotiation

protocol allows the agents to achieve better outcomes.

  

Luke Riley

 Talk: An Implementation of a Dialogue System for Collaborative Argumentation

 

Abstract: In this talk I will present recent work on an implemented system that enables autonomous agents  to engage in dialogues that involve inquiries embedded within a process of practical reasoning. The  implementation builds upon an existing formal model of value-based argumentation, which has itself been extended to permit a wider range of arguments to be expressed. I will discuss the formal extensions made to the work being built upon, the implementation, the interesting issues that have arisen from the implementation and the experimental avenues that can be pursued using this test-bed.

 

Tom van der Weide

 Talk: Argumentation for Decision-Support

 

Abstract: We consider two elements in supporting a user in decision-making: understanding what is the best for the user and effectively communicating this. Argumentation influences decision-making and is an effective means for decision-support.  To determine the best decision, the system needs to understand practical reasoning, which involves what the user values and how this translates to a decision situation. A value tree decomposes what a user values into general areas of concern, intermediate objectives and specific evaluation criteria. I have investigated the use of argumentation in practical reasoning using value trees.  To effectively communicate what decision is the best, it is common that an agent can select only one from many possible arguments. Multiple criteria typically influence what makes an argument effective for a user. Using our argumentation mechanism for value trees, a mechanism is proposed to reason about what argument is the most effective for a particular user.

 

Eric Kok

 Talk: Argumentation-based dialogues

 

Abstract: Where agents are employed to cooperatively solve tasks, communication becomes a critical factor. Argumentation-based models have been proposed to support the required dialogues. Improved efficiency and effectiveness of these dialogues have been the justification for using argumentation. However, as of yet there is little work done on making the benefits of arguing versus non-arguing models clear. Our current research focusses on empirically evaluating argumentation-based dialogues, specifically for deliberation, by quantifying the desirable properties for such dialogues and comparing arguing and non-arguing strategies in a multi-agent simulation.

 

Nabila Hadidi

 Talk: Tactics, Offers and Concessions for Argumentation based Negotiation

Abstract: Argumentation-based negotiation has gained increasing prominence in the multi-agent field over the last years. There is currently a long literature on the use of argumentation in negotiation and more especially on the modeling of negotiation protocols or decision making mechanisms. However the study of strategic issues that define the behavior of an agent during the negotiation has been largely neglected. This work fills this gap by providing, for the first time, several heuristics that implement profile, behavior and time based tactics that can be combined together to implement complex strategies similar to those studied in game-theoretic argumentation. Moreover, it presents several alternative methods for selecting the best offer to be proposed next, that are based on information about the preferences of the agents. The combination of these different tactics and offer selection methods lead to different types of concessions. Finally, an experimental evaluation, appearing for the first time in argumentation-based negotiation, shows how tactics, offers choices and concessions may influence the negotiation length and outcome, under the assumptions of time constraints and the availability of information on the opponent.

   

Dionysis Kontarinis

 Talk: Regulating Multiparty Persuasion with Bipolar Argumentation: Discussion and Examples

 

Abstract: The problem of regulating multiparty persuasion among several (more than two) self-interested agents has recently been put forward in a series of papers. Most of these works are settled in the original Dung’s framework of argumentation. In this paper, we discuss informally the challenges that arise when one wants to consider more expressive bipolar argumentation frameworks using a gradual evaluation of arguments. We see in particular that the notion of relevance can be interpreted in  different ways, giving rise to different types of protocols. The   discussion is illustrated by a number of examples.

SetPageWidth