Toward a Culture Adaptive Conversational Agent with a Modularized Approach

Embodied Conversational Agents (ECAs) are computer generated human-like characters that interact with human users in face-to-face conversations. ECAs are powerful tools in representing di fferences in cultural aspect and interacting with human users for training and edutainment applications. In this paper we present the preliminary results of our modularized approach by using our Generic ECA Framework to build a culture adaptive virtual tour guide agent for serving Japanese, Croatian and general western users with appropriate display of appropriate verbal and non-verbal behaviors. ACM Classification: H.1.2 [User /Machine Systems]: Human factors General terms: Design, Human Factors, Standardization.