COST Action IC0801

In4MAT - Infrastructures for Making AT

To add any new infrastructure to this page, send an e-mail with the subject “[COST-AT] Infrastructure” to carrasco@dsic.upv.es including the required information in plain text.

Framework/Infrastructure:

Obligations specification and monitoring using Semantic Web Technology.

Authors:

Nicoletta Fornara (University of Lugano, via G. Buffi 13, 6900 Lugano, Switzerland)

Marco Colombetti (Politecnico di Milano, piazza Leonardo Da Vinci 32, Milano, Italy)

E-Mail: {nicoletta.fornara, marco.colombetti}@usi.ch,

marco.colombetti@polimi.it

Link: http://www.people.lu.unisi.ch/fornaran/ObligationsOntology.html

Description:

In our past works we have proposed the OCeAN meta-model for the specification of artificial institutions and for the definition of the commitment-based semantics of an Agent Communication Language (ACL) [2, 3]. This meta-model can be used for the design of open interaction systems where agents may interact, negotiate, and reach agreements on their future actions.
Recently we started to specify the notion of obligation, which is fundamental in our model, using Semantic Web Technology[4,1]. In particular we specified the ontology for representing obligations, the state of the interaction, and for monitoring the evolution of the state of the obligations using OWL 2 DL3 and SWRL (Semantic Web Rule Language)4 rules, we used OWL-API 3.2.35 for updating the ontology from a Java program, and HermiT 1.3.46 for reasoning on the ontology.
The crucial characteristics of the proposed ontology are:

• It is possible to represent obligations whose content is related to time, with activation condition and deadline;
• The content of the obligations is a class of possible actions, one of these actions has to be performed within a given deadline in order to fulfill the obligation;
• The state of obligations is monitored using OWL axioms and a Java program that is necessary to be able to perform close world reasoning on certain classes. This because we assume that in the interaction contexts where this model will be used, not being able to infer that action has been performed in the past is sufficient evidence that the action has not been performed.

Framework/Infrastructure:

Case-Based Argumentation Infrastructure For Agent Societies

Authors:

Jaume Jordán, Stella Heras,  Vicente Julián

Departamento de Sistemas Informáticos y Computación

Universitat Politècnica de València

Valencia, Spain

E-Mail: {fjjordan,sheras,vingladag}@dsic.upv.es

Link: http://users.dsic.upv.es/~vinglada/docs/

Description:

Despite the important advances in the argumentation theory, it is difficult to find infrastructures of argumentation offering support for agents that have a social context. We propose an infrastructure to create and run argumentative agents in an open Multi-Agent System. This infrastructure offers the necessary tools to develop agents with argumentation capabilities, including the communication skills and the argumentation protocol, and it offers support for agent societies.

Argumentation theory has produced important benefits on many AI research areas, from its first uses as an alternative to formal logic for reasoning with incomplete and uncertain information to its applications in Multi-Agent Systems (MAS). Currently, the study of argumentation in this area has gained a growing interest. The reason behind is that having argumentation skills increases the agents’ autonomy and provides them with a more intelligent behaviour. An autonomous agent should be able to act and reason as an individual entity on the basis of its mental state (beliefs, desires, intentions, goals, etc.). As member of a MAS, an agent interacts with other agents whose goals could come into conflict with those of the agent. In addition, agents can have a social context that imposes dependency relations between them and preference orders among a set of potential values to promote/demote. For instance, an agent representing the manager of a company could prefer to promote the value of wealth  (to increase the economic benefits of the company) over the value of fairness  (to preserve the salaries of his employees). Therefore, agents must have the ability of reaching agreements that harmonise their mental states and that solve their conflicts with other agents by taking into account their social context. Argumentation is a natural way of reaching agreements between several parties with opposing positions about a particular issue. The argumentation techniques, hence, can be used to facilitate the agents’ autonomous reasoning and to specify interaction protocols between them.

In this work, we propose an infrastructure to create and run argumentative agents in a MAS. This infrastructure offers the necessary tools to develop agents with argumentation capabilities, including the communication skills and the argumentation protocol, and it offers support for agent societies. The main advantage of having this infrastructure is that it is possible to create agents with argumentation capabilities to resolve a specified problem. This approach can obtain better results than other distributed approaches due to the argumentation process between agents and their reasoning skills. In the argumentation dialogue the agents try to reach an agreement about the best solution to apply for each proposed problem.

Currently, the infrastructure has been implemented and tested in a real customer support application. All knowledge resources developed use ontologies as representation language. Concretely, we assume the existence of a domain-dependent ontology that represents the concepts of the application domain and we have created an argumentation ontology, called ArgCBROnto that agents that provides a common understanding about argumentation concepts for agents of agent societies. In addition, we have created OWL-API ontology parsers to manage the knowledge resources of our argumentation framework.

The main components of our infrastructure are the argumentative agents, the Commitment Store and the knowledge interchange mechanism. We consider different organizations or groups composed by some argumentative agents. The Commitment Store interacts with all the argumentative agents to store the positions and the arguments generated in the argumentation dialogue. The knowledge interchange is made with concepts of the ArgCBROnto ontology. The agent platform used in the implemented infrastructure is Magentix2. This is an agent platform that provides new services and tools that allow for the secure and optimised management of open MAS. The argumentative agents and the Commitment Store agent are extensions of the Magentix2 BaseAgent. Both the ontology and the parsers are now available, while the software to run the argumentation framework is being adapted to be able to use in generic problem solving applications and will be publicly available in a near future.

Framework/Infrastructure:

Agreement Technologies Environment (ATE)

Authors:

Carlos Carrascosa (UPV, Spain)

Marc Esteva (IIIA-CSIC, Spain)

Vicente Julián (UPV, Spain)

Marc Pujol (IIIA-CSIC, Spain)

Mario Rodrigo (UPV, Spain)

Juan A. Rodríguez-Aguilar (IIIA-CSIC, Spain)

Bruno Rosell i Gui (UDT-IA, IIIA-CSIC, Spain)

Matteo Vasirani (URJC, Spain)

E-Mail: rosell@iiia.csic.es

Link: http://www.iiia.csic.es/~rosell/ATE/

Description:

The Agreement Technologies Environment (ATE) must provide an environment where agents and/or humans can interact with each other to reach agreements.

This environment is aimed at supporting the establishment of coalitions, collaborations, negotiations and agreements on everything.

On the other hand, humans and / or agents participating in the ATE are not homogeneous, but each has its own objectives, their way of interacting, and can even be located in different locations. Namely the ATE should provide a heterogeneous environment where that humans as well as software agents can join.

Framework/Infrastructure:

Nüwa - A Web Service Directory

Authors:

Alberto Fernández, Carlos A. Soto, Zijie Cong

Universidad Rey Juan Carlos Calle Tulipán s/n, Móstoles, 28933, Spain

E-Mail: alberto.fernandez@urjc.es, casotob@ia.urjc.es, zijie.cong@urjc.es

Link: http://lovelace.escet.urjc.es/nuwa/

Description:

“Nüwa” is a directory for heterogeneous web service developed by Artificial Intelligence Group of Universidad Rey Juan Carlos. It addresses the issue of web service discovery involving heterogeneous description languages such as OWL-S, SAWSDL, WSDL and plain text.

Services description in different description languages are mapped into a unified service description model before registration. This unified model captures many important features of existing description languages, such as the semantic and syntactic I/Os, category information and syntactic description.

Matchmakers based on this unified model provide service discovery capabilities to Nüwa system. The currently available matchmaker performs semantic matching using domain ontology, syntactic matching using WordNet lexical database and categorization using NAICS 2007 classification system.

Combining Semantic Web technologies and Natural Language Processing techniques, Nüwa system intends to provide capability discovery ability for large, open multi- agent systems with higher precision and automation.

The design of the system is open and flexible, where new service description approaches and matchmakers can be easily integrated into it. This system provides a RESTful API interface for software agents and a web interface for human users.

As an ongoing project, the implementation is preliminary but sufficient for demonstrating the basic purpose of the project.

Framework/Infrastructure:

CArtAgO

Authors: A. Ricci, A. Santi

E-Mail: a.ricci@unibo.it

Link: http://cartago.sourceforge.net

Description:

CArtAgO is a framework and infrastructure for developing and running general-purpose distributed computational environments for multi-agent systems, designed in terms of artifacts as first-class abstraction defined by the A&A conceptual/meta-model [3].

The background idea is to have environments as first-class abstractions in MAS development [1] and - in particular - of environment programming [2], so exploiting the notion of environment not as a merely source of percepts and actions (typical AI view), but as an abstraction layer that can be designed and programmed to encapsulate functionalities and services that agents can share and use at runtime.

Such functionalities span from simply enabling and governing the access and interaction with the real external environment (being it sofware or hardware), hiding low-level details, to supporting the communication, coordination, and cooperation among agents, as well as their organisations.

In the case of CArtAgO, a (distributed) environment can be designed in terms of set of workspaces including a dynamic set of artifacts, representing tools and resources - programmed by the MAS developer - that the agents can dynamically create, use, compose and dispose as first-class entity of their world.

CArtAgO is orthogonal with respect to the specific agent model adopted [5], so it can be integrated and used by agents possibly based on heterogeneous computation model & architecture, so as to develop open multi-agent systems possibly composed by heterogeneous agents sharing and interacting inside the same artifact-based environments. In spite of this orthogonality, the A&A model and CArtAgO it have been conceived to be particularly effective when integrated with cognitive agent languages/frameworks/platforms, such as the one based on the BDI model/architecture.

CArtAgO technology includes [4]:

• a Java based API to program the environments
• a runtime / infrastructure to run the environments, possibly distributed over the network
• bridges that allow for integrating CArtAgO with different kinds of agent programming languages/framwork/platforms. The most updated one is c4jason bridge - which is the core of a family of platforms based on the tight integration of Jason and CArtAgO (JaCa, JaCaMo, JaCa-Android). Other bridges, still to be updated for the most recent version of CArtAgO, include bridges to Jadex and AgentFactory. Besides these ones, a raw Java API is provided to build custom bridges to what ever existing agent platform.

[1] D. Weyns, A. Omicini, J. J. Odell, Environment as a first-class abstrac- tion in multi-agent systems, Autonomous Agents and Multi-Agent Systems 14 (1) (2007) 5-30.
[2] A. Ricci, M. Piunti, and M. Viroli. Environment programming in multi-agent systems: an artifact-based perspective. Autonomous Agents and Multi-Agent Systems, 23:158-192, 2011
[3] A. Omicini, A. Ricci, M. Viroli, Artifacts in the A&A meta-model for multi- agent systems, Autonomous Agents and Multi-Agent Systems 17 (3) (2008) 432-456.
[4] A. Ricci, M. Piunti, M. Viroli, A. Omicini, Environment programming in CArtAgO, in: R. H. Bordini, M. Dastani, J. Dix, A. El Fallah-Seghrouchni (Eds.), Multi-Agent Programming: Languages, Platforms and Applica- tions, Vol. 2, Springer, 2009, pp. 259-288.
[5] A. Ricci, M. Piunti, L. D. Acay, R. Bordini, J. Hübner, and M. Dastani. Integrating artifact-based environments with heterogeneous agent-programming platforms. In Proceedings of 7th International Conference on Agents and Multi Agents Systems (AAMAS08), 2008.

Framework/Infrastructure:

JaCa-Android

Authors: A. Santi, A. Ricci

E-Mail: a.ricci@unibo.it

Link: http://jaca-android.sourceforge.net

Description:

Platform for developing agent-based applications and multi-agent systems based on the JaCa model (Jason+CArtAgO) on the top of Android Mobile computing platform [1].

[1] Andrea Santi, Guidi Marco, Alessandro Ricci. JaCa-Android: An Agent-based Platform for Building Smart Mobile Applications. In Proceedings of LAnguages, methodologies and Development tools for multi-agent systemS (LADS-2010), 2010.

Framework/Infrastructure:

JaCaMo

Authors: O. Boissier, R. Bordini, J. Hubner, A. Ricci, A. Santi

E-Mail: a.ricci@unibo.it

Link: http://jacamo.sourceforge.net

Description:

General-purpose platform for developing multi-agent systems based on Jason+CArtAgO+Moise providing an integrated approach to Multi-Agent Programming, including main programming dimensions such as Agent-Oriented Programming, Organisation-Oriented Programming and Environment Programming.