Toward a Multi-culture Adaptive Virtual Tour Guide Agent with a Modularized Approach

Embodied Conversational Agents (ECAs) are computer generated humanlike characters that interact with human users in face-to-face conversations. ECA is a powerful tool in representing the differences in cultural aspects and suitable for interactive training or edutainment systems. This paper presents the preliminary results of the development of a culture adaptive virtual tour guide agent for serving Japanese, Croatian and general western users by displaying of appropriate verbal and non-verbal behaviors. It is being implemented in Generic ECA Framework, a modulized programming framework for developing ECAs. By dividing ECA functions to reusable and loosely coupled modules, minimum efforts are required and it is supposed easy to incrementally scale up the system.