Cerri, S.A., Maraschi, D. The relations between Technologies for Human Learning and Agents; AFIA 2001; Atelier: Méthodologies et Environnements pour les Systèmes Multi-Agents ; LEIBNIZ-IMAG, Grenoble, 25-29 juin 2001 page 1 of 11 ____________________________________________________________________ The relations between Technologies for Human Learning and Agents Daniele Maraschi and Stefano A. Cerri LIRMM, Rue Ada 161, 34392 MONTPELLIER CEDEX 5, France {cerri,maraschi}@lirmm.fr http://www.lirmm.fr Abstract. In this position paper we review the historical emergence of Agents and the current developments around our integrated Web and Agent platform. The focus of the paper is to show that research results in Intelligent Tutoring and Learning Environments are strongly related to most of current hot issues in Agents and Agent Communication Languages, and vice-versa: that realistic dialogues for Human Learning will be generated by advances in models and tools for Agent Communication. Keywords: Agent Communication Languages, Interaction Languages, Web Languages, Technologies for Human Learning, Software Engineering, Distributed Systems, Functional Programming, Object Oriented Programming. 1. The historical emergence of Agents From the seminal work of Newell concerning the Knowledge Level in Knowledge Based Systems [1], we know how to separate the analysis and synthesis of Knowledge from the ways it may be implemented at the Symbol level. Different approaches may be adopted, e.g. Description Languages that assume Ontologies and identify methods for deducing Facts, Relations and Rules from interoperable multiple descriptions. An orthogonal approach based on Propositionalisation, instead, uses Terms, Constraints and Machine Learning for the same purposes [2]. As Computing is the art of transforming semantics into syntax, evaluating results of syntactic processing, and mapping them back to semantics; the historical challenge in Computing consisted of relating syntactic structures to their semantics with respect to meanings in the real world. Semantics denotes here the real world concepts, not just the "world of the Computer" at in the case of Denotational, Operational or Algebraic semantics of programming languages. Due to the availability of the Web, the process is currently more and more conversational, between and among "abstract computational entities", human or artificial ones, called Agents, on the Web. As a consequence, people are more concerned with conversations on the Web, their semantics 1 , their pragmatics. Among existing conversations, and those that are envisioned to occur in the next future with an impressive growth rate, we consider commercial transactions in e- commerce and intellectual transactions in e-learning to be pivotal for crucial developments. We also assume as intuitive the idea that e-learning and e- commerce, as well as, in general, e-work conversations, share a similar nature and therefore are characterized by similar requirements. 1 See, for instance: http://www.semanticweb.org/SWWS where our impression is that people that did not succeed to agree about standard ontologies on single machines, in order to build complex knowledge-based systems, try to export the same approach to the Web. The very first confusion arises from the attribute "semantic": is the Web supposed to become "semantic" or the "emergent semantics of the Web" will be generated - attributed by humans?