Embodied conversational agents are synthetic characters that resemble humans not only in their physical appearance, but also in their behaviour with more or less the communicative characteristics of real persons. They are gaining much interest, both as mediums to visualize models of human communication, and as interfaces in different applications. The first steps have been taken in the direction of expressive ECAs, by endowing them with the capability of showing emotions.

Subtle issues such as the impact of social role and personality have also been addressed. Non-verbal signals have been used to accompany speech to make a virtual human more expressive and believable, but, these works concentrated on modeling the psychological, social and communicative aspects of emotional and cognitive state. Usually the presentational issues are not dealt with as a research topic, but as a practical task for an animator, often only to make a specific application or demonstrator. Moreover, most of the agents created so far are generic: it is difficult to differentiate them by the way they move, answer, react? It can be said that most of the time, they all "come from the one mould".

This workshop will bring both established and new researchers together to discuss the development, implementation and evaluation of ECAs as individuals , the current ECA state-of-the-art and to create a vision for the future of individual ECAs.

The origin of this workshop dates from a successful workshop on Embodied conversational agents - let's specify and evaluate them! for AAMAS 2002, Bologna in mid July 2002 and a special IST 5^th Framework meeting on Representation Formats/Languages.

• identification of the factors influencing expressiveness and individuality; the factors may come from linguistics, speech, emotion, personality, gesture, social relation and/or dialogue capabilities.
• description, parameterisation, representation of such values;
• evaluation of a particular representation of individualism - what are the useful features; what are the limitations; in how far are original requirements met ?

• description of the parameters defining the expressiveness/individuality of an agent with a brief description of application area, including the type of agent(s) to be achieved and the research goals.
• pointers, such as on-line demos or videos, to actual working examples that demonstrate the creation of such agent

The workshop will conclude with a panel discussion of:

• Important issues raised within presentations,
• A visionary glimpse of / plan for the future for ECAs as individuals.

We strongly encourage you to demonstrate your work and to provide URLs to permanent web sites that contain examples of your system.

To give us time to organize the technical requirements, if you are to demonstrate your system, we must be notified before 15^th April . Contact us for details at: raytrace@vhml.org

At the end of the workshop, participants should have a broad understanding of the differing technologies involved in the production, development and directing of "individualised ECAs". They will have seen several ECA systems and will understand the problems associated with the technology. They will have contributed, through discussion, to the future development of individual ECA interfaces within the global context. They will also become part of a network of ECA researchers and practitioners.