Integrating embodied conversational agent components with a generic framework

Embodied Conversational Agents (ECAs) are computer generated life-like characters that interact with human users in face-to-face conversations. To achieve natural multi-modal conversations, ECA systems are sophisticated and require numbers of building assemblies. They are thus difficult for an individual research group to develop. To address this problem, we are developing an approach to connect those components with a Generic ECA (GECA) framework. GECA is composed with a blackboard-model based platform, a high-level protocol and a set of APIs which are meant for easing component wrapper development. As an expectation, with such a generic ECA framework, rapid ECA system prototyping is possible while research result sharing and the collaboration between ECA researchers can be facilitated. This paper describes the basic concepts of this framework, an initial implementation and evaluations of actually using it to build a realistic ECA application.

• Type: Journal
• Year: 2008
• Where: MAGS
• Authors: Hung-Hsuan Huang, Aleksandra Cerekovic, Kateryna Tarasenko, Vjekoslav Levacic, Goranka Zoric, Igor S. Pandzic, Yukiko I. Nakano, Toyoaki Nishida
• Keywords: ECA Systems, Generic Eca, Generic ECA Framework, MAGS 2008