- Date & Time : 2011. Dec. 15 (thursday), 10:00 A.M
- Place : Seminar Room (#1222)
- Speaker : Gahgene Gweon (권가진)
- My research focuses on building systems to support the efficacy of group functioning using natural language processing and machine learning. What is powerful about my research is that I use technologies in ways that are informed by insights from the behavioral sciences. My research program is organized along two axes: operationalizing processes that support collaborative work (behavioral science) and automatically monitoring processes that support collaborative work (design & computer science). In this talk i will present two research studies in each of the axis.
The first study was conducted in an educational setting to show that environments that provide more opportunities to students for reflection yield learning benefits. More specifically, I experimentally compared dynamic support vs. no support (Gweon, et al., 2006), in tutoring systems that were designed and built to test theories of collaborative learning. Positive results from the first study led to multiple systems that provide dynamic support for collaborative learning systems. It also served as an opportunity to bridge the two communities of computer supported collaborative learning (CSCL) and intelligent tutoring systems (ITS), by allowing technologies that were available in ITS to be applied to the CSCL theory.
In the second study, I present my thesis work which examines how students work in groups, with the goal of creating technology to monitor group processes. More specifically, I use speech as a data source to detect a specific type of knowledge sharing process, namely the idea co-construction process (ICC). I propose to detect ICC without using speech content, but using just the prosody of speech. The grounds for my proposal is based on the literature from sociolinguistics. Sociolinguists have shown for decades now, that there are social cues present in speech prosody. The proposal of this hypothesis is innovative in that I brought the cutting edge technology using machine learning and speech feature extraction techniques from the area of language technologies and applied those technologies to the theories in sociolinguistics.