The Articulab’s SCIPR project aims to understand how to foster curiosity in students by observing them in small group settings. I am taking part in the development of an embodied conversational agent (ECA), in our study an intelligent virtual child, to engage in a collaborative tabletop game, and elicit curiosity during the game play. The virtual child’s behavior will be integrated as part of a WoZ system, which allows the above behaviors in a semi-automated matter. Find more about my experience here. Working under PI Dr. Justine Cassell, supervisor Dr. Zhen Bai.
For this project, I took part in a very iterative design process that involved observation of audio/video data, transcription, coding schemes, creating finite state machines, and frequent play testing. In addition to this, I used several NLP techniques to extract and analyze speech patterns of real children in order to build finite state machines for an AI system. The exploratory work I did as this project gathers steam lays the groundwork for the AI system that will interact with real children in the future.