Towards a Artificial Social Intelligence: Autonomous Agents and their Social Order

Prof.dr. Cristiano Castelfranchi , Institute for Cognitive Science and Technology - CNR, Roma

I will illustrate how and why AI - reacting to strong attacks and criticisms and to a serious crisis - has changed its paradigm, moving from an 'isolated', 'reasoning', 'complete' view of intelligence to a 'social' paradigm were computation mainly is 'communication' and problem solving is due to (emerging) cooperation between decentralized, autonomous, situated, interactive, limited intelligent entities (Agents).
I will analyze the relational and social nature of 'autonomy', how it depends on the cognitive architecture of the agent and on its internal and external powers (like knowledge or material resources and authorizations). What are the necessity and the advantages of Autonomy, and its problems: like collaborative conflicts (for example in user-Agent interaction).
I will discuss how this social paradigm is shaping the entire discipline: the 'social' nature of information and of IT. Why social notions and capacities (like cooperation, negotiation, role, commitments, norms, Š) must be computationalized in order: a) computers (and agents) can really mediate and support human cooperation; c) H-C and H-robot interaction be natural and effective; c) Multi-Agent System can really work producing market-like, organization-like, institution-like form of computing and various forms of spontaneous or organized 'social order'.

Reasoning about Knowledge, Rationality and Action

Prof. dr.Wiebe van der Hoek, Universiteit van Liverpool

Since the work of Hintikka on Logic for Knowledge, researchers from Computer Science, Artificial Intelligence and Multi-Agent Systems have appreciated and added to this formal theory, in which one can reason about how agents (participants in a communication protocol, or in a secret exchange, or in a game like scenario) are guaranteed to either obtain knowledge or remain ignorant. We will briefly present the basics of this logic, in which in particular the Multi-Agent notions of knowledge are intriguing, like common knowledge as illustrated in the muddy children example.
An important issue of such scenarios that is only recently formally analyzed, however, is that of the dynamics. How do the agents update their knowledge, and how does their newly acquired knowledge influence their decision making? We will sketch two approaches that address the dynamics of epistemics, in which we can specify formally the muddy children scenario, as well as automatically verify epistemic properties in certain knowledge-exchange scenarios.

Domain Modelling; Anchoring semantics in information systems development

Prof. dr. Erik Proper, Katholieke Universiteit Nijmegen

This presentation focuses on domain modelling and its role in the development of information & knowledge systems. We define domain modelling as the activities involved in obtaining and modelling the language (concepts, terminologies; ontologies) used by stakeholders to discourse about a domain.
Achieving conceptual clarity and consensus among stakeholders is an important yet often neglected part of modelling activities in system development. During system development, a myriad of models may be produced, ranging from high level sketches of the problem, via informal/formal requirements, to designs at several levels of technical detail. Underlying each of these models, a domain model can be discerned, comprising the concepts (domain concepts) featuring in the model. The different models used during system development need to be communicated with different stakeholders, ranging from problem owners, contract authorities, perspective users, domain experts, to software engineers and system administrators. This puts an extra burden on task of domain modelling as a shared understanding must be reached of the concepts involved.
In this presentation we will investigate the importance of domain modelling and its role during system development. Based on this we will pose some challenges that we regard as essential in improving the awareness for and practice of domain modelling in a system development context.

The making of a Web Ontology Language: a chair's perspective

Prof. dr. Guus Schreiber, Vrije Universiteit Amsterdam

Abstract
In the context of the Semantic Web Activity the World-Wide Web Consortium (W3C) installed in 2001 a new working group to define a web standard for representing ontologies. The discussions in this group reflects a long history of debates in knowledge representation and knowledge engineering. In this talk I will discuss the process that led to the Web Ontology Language OWL, both from a technical and social point of view. The technical focus will be on the compromises between expressivity and computability and on the relationship with RDF. Social aspects include observations on strengths and weaknesses of the W3C standardization process and on the art of consensus. Forthcoming W3C semantic-web activities will be briefly outlined.