The pragmatic of software agents: analysis and design of agent communication languages

 
EPrints.org
Agentlink Clearinghouse is powered by GNU EPrints developed by the School of Electronics and Computer Science of the University of Southampton.
Type: Thesis (PhD)
Member Organisation: 029 Universidad Rey Juan Carlos

Serrano, J. M. (2004) The pragmatic of software agents: analysis and design of agent communication languages. PhD thesis, University Rey Juan Carlos.

This is the latest version of this eprint.

Full text not available from this archive.

Abstract

This thesis focuses on several conceptual, methodological and implementation concerns of agent communication language pragmatics, namely communicative actions and protocols. The research is carried out under two major assumptions. Firstly, the pragmatics of software agents is an essential feature in the definition of agenthood. Secondly, both the specification and instrumentation of multi-agent systems take place in a broader organizational perspective. The thesis generally aims at bridging the gap between this two major dimensions of multi-agent systems. The research was initially motivated by the problem of speech act classification -- an issue first addressed by philosophers J. L. Austin and J. R. Searle. In constrast to the common approach to this problem in the philosophy of language, we put forward an organizational stance on agent communicative language pragmatics. Essentially, we argue that catalogues of communicative actions and protocols may be structured in terms of a collection of generic roles and interations. The approach is formalized in terms of the so-called RICA meta-model and exemplified by the structuring of the FIPA ACL catalogues. The benefits of this perspective are first shown by specifying a method of agent communication language design, which allows to smoothly integrate in a principled-way the communicative features of the multi-agent system within its organizational model. It further shows how generic communicative roles and interactions shape the interaction space of organizational models. Moreover, the method properly handles the trade-off between reusability and expressiveness by allowing the extension of standard catalogues in a structured way. A novel technique for interaction protocol design, based on so-called Interaction State Machines, complements the design method. It illustrates how interaction protocols may be systematically designed in an organizational context, and the advantages of recursive definitions. Finally, the thesis attempts to draw the major conclusions at the implementation level, by instrumenting the RICA metamodel and the Interaction State Machine formalism on top of the Jade platform. On the one hand, the resulting framework, so-called RICA-J(ade), allows multi-agent system designers to instrument the organizational model of the application by customising the software components available in organizationally-structured libraries of reusable agent communication dialects. On the other hand, independent agent programmers may then rely on the resulting organizational library, which may be adapted to their particular needs. As a major advantage, the RICA-J framework relieves agent programmers from the instrumentation of generic issues concerning the management of interactions, and puts forward a component-based approach to the instrumentation of multi-agent organizations. Two sample prototypes within an e-commerce and a conference management domain serve to illustrate the proposal.

Deposited by Juan Manuel Serrano on 14 April 2005

Available Versions of this Item

Archive Staff Only: edit this record

   

AgentLink is the European Commission's IST-funded Coordination Action for Agent-Based Computing
and is coordinated by the
University of Liverpool and University of Southampton
If you encounter any problems with these pages please contact web@agentlink.org.