Case-based Argumentation Infrastructure for Agent ... - Semantic Scholar
Recommend Documents
move mv0 since none faculty has priority over fa1. At the end of the game, Pro contains the assumptions of an admissible argument deducing the motiv(ag ),.
claim, and an attack relation, i.e., a directed relation that represents conflict between ... gic exploitation, in particular, when agents have knowl- edge about their ..... (assuming aliens can be regarded as living beings). For some discussion on .
Logic Programming (DeLP) by Garcıa and Simari where agents are engaged in an argumentation to ... As the underlying logical foundation we use Defeasible Logic Programming (DeLP). [13] which is a ...... pages 381â392. IOS Press, 2008.
Email: {zhljenny, parashar}@caip.rutgers.edu. September 26, 2005. Abstract ..... Figure 5: Example of tuple retrieval in Comet: (a) the template defines a.
research. KQML and the FIPA ACL are the leading can- ... the latest FIPA draft standard [16, 12]. ... described so that they are accessible both to people and to.
Sep 26, 2005 - Autonomic development and management strategies, which are inspired by bi- ological .... Such an infrastructure has to implement services.
Keywords: Enterprise integration, knowledge engineering, distributed ..... call for bids, waits for a reply for some length of time, and then awards a contract to the best .... considered as the administration center of the manufacturing enterprise.
Dept. of Computer Science. National ... Of course, agent can verify each other's identity certificate under ... except RA on the top of APKI who registers to its MA.
Dung argumentation framework is itself instantiated by arguments that make state- ... then be uniformly characterised by metalevel argumentation in a Dung ...
Jun 17, 2011 - In computing, large systems can be viewed or designed in terms of the services they offer and the entities that interact to provide or consume ...
Complexity Results for Argumentation-based Agent Communication. Jamal Bentahar1 and Zakaria Maamar2. 1CIISE, Concordia University, Montreal, Canada ...
In other words agents will only accept things if they don't have a good reason not to. ..... Should an agent only be able to argue what it believes to be true? If not,.
We add EA6 to the framework, as shown in Figure 2 (arrow 3). EA3 is attacked by an argument with the same value and so is defeated. If safety is preferred to ...
Tools are needed to analyse firewall configurations .... illustrator which allows a graphical representation of the rule- set. ... This can lead to the introduction of a.
365 â 384. ISSN 1380-3611 (print)/ISSN 1744-4187 (online)/05/040365â20 ... Wegerif, and Dawes (1999) on student-student interactions support the importance ...... Confronting cognitions in computer-supported collaborative learning.
consider the following set of arguments: (a) The bridge should be built on solid ... 1Correspondence to: Nir Oren, Department of Computer Science, King's ...
Abstract. Argumentation is important for agent negotiation. In this paper, we develop an efficient framework for multi-agent argumentation. We identify as-.
Dec 18, 2005 - harder to ignore when the map is assessed at decision time. ..... Available from stadium.open.ac.uk/stadia/preview.php?s=29&whichevent=494.
Computer-Supported Argumentation for Cooperative Design ..... 365-387. 20] Trousse, B., Visser, W.: Use of Case-Based Reasoning Techniques for Intelligent ...
Computational Models of the Law. 2. Motivation. Assumption-based argumentation can serve as an effective computational tool for argumentation-based ...
to have blind faith in his agent's ability to trust other agents. We believe that the state of the art in dealing with trust in Multi-Agent Systems has not suf-.
After formalising the concept and presenting a simple argumentation framework in which it can be used, we show a sample dialogue utilising the heuristic. We.
Second) of contracts have been proposed to ârestateâ clearly and precisely the principles and rules of common law[21]. Thus modeling legal doctrines offers an ...
Argumentation-based agents for eProcurement. (Short Paper). Paul-Amaury MATT and Francesca TONI. Imperial College London. Department of Computing.
Case-based Argumentation Infrastructure for Agent ... - Semantic Scholar
It is important to offer support for agent societies and their agents' social context. ... CBR methodology has been mostly used to generate, select or evaluate ...
Case-based Argumentation Infrastructure for Agent Societies Jaume Jordán Departamento de Sistemas Informáticos y Computación Universitat Politècnica de València
LAF 2012
1 / 23
Outline 1
Introduction Motivation Proposal State of the Art
2
Infrastructure Argumentative agents Commitment Store Knowledge interchange mechanism
3
Call Centre Example Description Argumentation process Results
2 / 23
Introduction
Motivation
Motivation
Argumentation theory has produced important benefits on its applications in Multi-Agent Systems (MAS). Argumentation skills increase the agents’ autonomy and provide them with a more intelligent behaviour. Agents must have the ability of reaching agreements to solve their conflicts. It is important to offer support for agent societies and their agents’ social context. It allows to simulate different models of organizations and societies to solve problems.
3 / 23
Introduction
Proposal
Proposal
We propose an infrastructure to develop and execute argumentative agents in an open MAS. It offers the necessary components to develop agents in an open MAS with argumentation capabilities, including the communication skills and the argumentation protocol. It offers support for agent societies and their agents’ social context. Agents use Case-Based Reasoning (CBR) to store domain knowledge of past solved problems and to store argumentation knowledge of previous dialogues. Agents try to reach an agreement about the best solution to apply for each proposed problem.
4 / 23
Introduction
Proposal
Open Multi-Agent Systems Agents can enter or leave the system, interact and dynamically form groups to solve problems. Provides great flexibility and diversity to the system. Open communications due to the heterogeneity of the agents. Establish control and security agents. Agents acquire a role for which a set of rules are established to control their behaviour.
5 / 23
Introduction
Proposal
Open Multi-Agent Systems Agents can enter or leave the system, interact and dynamically form groups to solve problems. Provides great flexibility and diversity to the system. Open communications due to the heterogeneity of the agents. Establish control and security agents. Agents acquire a role for which a set of rules are established to control their behaviour. Magentix2: agent platform that provides new services and tools to manage open MAS. THOMAS: open MAS framework based on modular web services to manage organizations. The OMS controls the organizations and the SF controls the services. Integrated in Magentix2. 5 / 23
Introduction
Proposal
Case-Based Reasoning Solved problems of past situations represented as cases are stored in a database called case-base.
6 / 23
Introduction
State of the Art
CBR Applications of Argumentation in AI
The PERSUADER system: a mediator agent centralises the negotiation process between two parts. Negotiation. CBR for Argumentation with Multiple Points of View: is centred on the argument diagramming for humans, helping agents to follow the discussion and supporting them with tools to pose better arguments. Deliberation. Case-based Negotiation Model for Reflective Agents: all agents are autonomous and able to start and manage a direct dialogue with other agents. Negotiation. Argument-based selection Model (ProCLAIM): deals with the internal deliberation of the mediator agent, supporting only this agent to make the best decision among the set of potential winners. Deliberation. Argumentation-based Multi-Agent Learning (AMAL): all agents decide the best classification tag for a specific object and cooperate by aggregating their knowledge in the deliberation process. Deliberation.
7 / 23
Introduction
State of the Art
CBR Applications of Argumentation in AI
CBR methodology has been mostly used to generate, select or evaluate arguments considering previous similar experiences. All frameworks are domain specific. None of them offers support for agent societies and their agents’ social context. Not suitable for open MAS.
8 / 23
Infrastructure
Argumentation framework
Agent society: agents that play a set of roles. Dependency relation between roles: charity, authorisation, power. Groups: Collaborate and reach the global objectives. Promoted values by agents and group.
Knowledge resources: Domain cases: represent previous problems and their solutions. Domain-dependent. Argument cases: represent previous arguments and their final outcome.
Arguments: To support a position (solution). To attack a different position.
9 / 23
Infrastructure
Infrastructure diagram
10 / 23
Infrastructure
Argumentative agents
Argumentative agents Argumentative agents Agents with argumentation capabilities that engage in an argumentation dialogue to reach an agreement about the best solution to apply to a problem.
11 / 23
Infrastructure
Argumentative agents
Argumentative agents Argumentative agents Agents with argumentation capabilities that engage in an argumentation dialogue to reach an agreement about the best solution to apply to a problem. Domain CBR Stores domain knowledge of previous solved problems.
11 / 23
Infrastructure
Argumentative agents
Argumentative agents Argumentative agents Agents with argumentation capabilities that engage in an argumentation dialogue to reach an agreement about the best solution to apply to a problem. Domain CBR Stores domain knowledge of previous solved problems. Argumentation CBR Stores ences.
past
argumentation
experi-
11 / 23
Infrastructure
Argumentative agents
Argumentative agents Argumentative agents Agents with argumentation capabilities that engage in an argumentation dialogue to reach an agreement about the best solution to apply to a problem. Domain CBR Stores domain knowledge of previous solved problems. Argumentation CBR Stores ences.
past
argumentation
experi-
Argument management process How the positions, support arguments and attack arguments are generated by the agents using their knowledge resources and their CBRs. It also defines the argumentation mechanism. 11 / 23
Infrastructure
Argumentative agents
Domain-Case structure
Domain-Case
12 / 23
Infrastructure
Argumentative agents
Argument-Case structure
Argument-Case ARGUMENT-CASE
Problem
Domain Context Premises
Proponent ID Role Value Preference Relation
Solution
Social Context
Solution Applied Value Promoted Acceptability Status Received Attacks: - Counter-examples - Distinguishing premises
Dependency Relation
Opponent ID Role Value Preference Relation
Group ID Role Value Preference Relation
13 / 23
Infrastructure
Commitment Store
Commitment Store Resource that stores all the information about the agents participating in the problem-solving process. Argumentation dialogues, positions and arguments. Every agent can read the information of the dialogues that it is involved in. It has been implemented as an agent to allow a good communication with the other agents.
14 / 23
Infrastructure
Knowledge interchange mechanism
Knowledge interchange mechanism The case-bases of the domain CBR and the argumentation CBR are stored as OWL 2 data. We have designed an ontology that acts as language representation of the cases. Heterogeneous agents can use it as common language to interchange solutions and arguments generated from the case-bases of the argumentation framework.
15 / 23
Infrastructure
Knowledge interchange mechanism
Argumentation protocol
16 / 23
Infrastructure
Knowledge interchange mechanism
Messages interchange
Figure: Messages interchange between argumentative agents
17 / 23
Call Centre Example
Description
Call Centre Problem
A Call Centre provides technological and customer support services. The operators attend to customer requests by means of a Call Centre. Typically organised in 3 levels: Base operators. Expert operators. Managers.
The Call Centre receives the request and creates a new incidence (Ticket). The operators work cooperatively to solve the Ticket. We represent the operators by means of argumentative agents.
18 / 23
Call Centre Example
Argumentation process
Argumentation process
1 OPERATORS
3
5
4 MANAGER
1 5
DOMAIN CASES
2 6
OWL PARSER
1 5
2
5
DOMAIN-CBR
DOMAIN CASES
DOMAIN-CBR
4 2 4
2 OPERATOR 1
OPERATOR 2
6 ARGUMENT CASES
OWL PARSER
6
4 6
OWL PARSER
OWL PARSER
ARGUMENT CASES
3
3
ARG-CBR
ARG-CBR CS
MAGENTIX2 PLATFORM
Figure: Data-flow for the argumentation process of the Call Centre application
19 / 23
Call Centre Example
Argumentation process
Argumentation process 1
The initiator agent sends the ticket to the agents.
2
Each agent evaluates if it can engage in the dialogue offering a solution. Queries to the Domain-CBR and Argumentation-CBR to select the best position (solution) to defend.
3
Positions defended by agents are stored by the CS.
4
Argumentation process: agents try to defend their positions and attack other different positions.
5
The dialogue is stopped after a specified time or when nothing new is proposed after a specific time. The most frequent position will be the solution.
6
Each agent updates its Domain-CBR and its Argumentation-CBR case-bases.
20 / 23
Call Centre Example
Argumentation process
Argumentation process 1
The initiator agent sends the ticket to the agents.
2
Each agent evaluates if it can engage in the dialogue offering a solution. Queries to the Domain-CBR and Argumentation-CBR to select the best position (solution) to defend.
3
Positions defended by agents are stored by the CS.
4
Argumentation process: agents try to defend their positions and attack other different positions.
5
The dialogue is stopped after a specified time or when nothing new is proposed after a specific time. The most frequent position will be the solution.
6
Each agent updates its Domain-CBR and its Argumentation-CBR case-bases.
20 / 23
Call Centre Example
Argumentation process
Argumentation process 1
The initiator agent sends the ticket to the agents.
2
Each agent evaluates if it can engage in the dialogue offering a solution. Queries to the Domain-CBR and Argumentation-CBR to select the best position (solution) to defend.
3
Positions defended by agents are stored by the CS.
4
Argumentation process: agents try to defend their positions and attack other different positions.
5
The dialogue is stopped after a specified time or when nothing new is proposed after a specific time. The most frequent position will be the solution.
6
Each agent updates its Domain-CBR and its Argumentation-CBR case-bases.
20 / 23
Call Centre Example
Argumentation process
Argumentation process 1
The initiator agent sends the ticket to the agents.
2
Each agent evaluates if it can engage in the dialogue offering a solution. Queries to the Domain-CBR and Argumentation-CBR to select the best position (solution) to defend.
3
Positions defended by agents are stored by the CS.
4
Argumentation process: agents try to defend their positions and attack other different positions.
5
The dialogue is stopped after a specified time or when nothing new is proposed after a specific time. The most frequent position will be the solution.
6
Each agent updates its Domain-CBR and its Argumentation-CBR case-bases.
20 / 23
Call Centre Example
Argumentation process
Argumentation process 1
The initiator agent sends the ticket to the agents.
2
Each agent evaluates if it can engage in the dialogue offering a solution. Queries to the Domain-CBR and Argumentation-CBR to select the best position (solution) to defend.
3
Positions defended by agents are stored by the CS.
4
Argumentation process: agents try to defend their positions and attack other different positions.
5
The dialogue is stopped after a specified time or when nothing new is proposed after a specific time. The most frequent position will be the solution.
6
Each agent updates its Domain-CBR and its Argumentation-CBR case-bases.
20 / 23
Call Centre Example
Argumentation process
Argumentation process 1
The initiator agent sends the ticket to the agents.
2
Each agent evaluates if it can engage in the dialogue offering a solution. Queries to the Domain-CBR and Argumentation-CBR to select the best position (solution) to defend.
3
Positions defended by agents are stored by the CS.
4
Argumentation process: agents try to defend their positions and attack other different positions.
5
The dialogue is stopped after a specified time or when nothing new is proposed after a specific time. The most frequent position will be the solution.
6
Each agent updates its Domain-CBR and its Argumentation-CBR case-bases.
20 / 23
Call Centre Example
Results
Prediction error Testing policies: CBR-Random (CBR-R): choose randomly a solution of domain-cases case-base proposed by the agents, without using argumentation. CBR-Majority (CBR-M): choose the most frequently solution of domain-cases case-base proposed by the agents, without using argumentation. CBR-Argumentation (CBR-A): agents perform an argumentation dialogue to select the best solution of those proposed.
Figure: Mean locutions generated per dialogue and prediction error
22 / 23
Summary
Summary Design and implementation of an infrastructure to develop and execute argumentative agents and all the components needed. The agents’ logic and their argumentation protocol. An ontology using OWL 2 to represent all the needed knowledge to facilitate the understanding between heterogeneous agents. A knowledge interchange mechanism using FIPA-ACL messages to communicate the agents. Two different CBR architectures for the domain CBR and the argumentation CBR.
Validation and evaluation with a real example, obtaining better performance than other reasoning approaches without argumentation. Future work: Our infrastructure will be used in other domains to measure the performance and the differences having a largest database of solved problems. 23 / 23
