FacebookTwitterGoogle PlusRSS News Feed

LTI Thesis Proposal

Thesis Proposals
Language Technologies Institute
Carnegie Mellon University
Agile Facilitation for Collaborative Learning
Wednesday, February 26, 2014 - 4:00pm
6115 
Gates&Hillman Centers
Abstract:

This proposed work investigates the use of adaptable conversational agents to scaffold on-line collaborative learning discussions through an approach called Academically Productive Talk (APT), or Accountable Talk. In contrast to past work on dynamic support for collaborative learning, where agents were used to elevate conceptual depth by leading students through directed lines of reasoning, this APT-based approach uses generic prompts that encourage students to articulate and elaborate their own lines of reasoning, and to challenge and extend the reasoning of their teammates.

Our body of completed work integrates findings from a series of studies across content domains (biology, chemistry, engineering design), grade levels (high school, undergraduate), and facilitation strategies. The pattern of results demonstrates that APT based support for collaborative learning can significantly increase learning, but that the effect of specific APT facilitation strategies is context-specific. It appears the effectiveness of each strategy depends upon factors such as the difficulty of the material and the skill level of the learner. As a primary contribution of this work, we plan to develop and operationalize a framework for automated, context-responsive facilitation. Through a series of in-vivo experiments and transcript analyses, we seek to verify and clarify which contextual factors are most critical when selecting suitable facilitation strategies.

Prior work has shown that models learned from sequences of discourse acts can predict learning and social outcomes, even from just the early stages of an ongoing discussion, and that representations of a conversation's evolving state hold benefit for both human and automated support. How do differences among macro-level collaborative contexts and turn-level interventions affect a group's conversational trajectory? Using discourse analysis methods, sequential modeling, and text classification, we will explore the potential of modeling conversational context and trajectory to select the most appropriate strategies for collaborative facilitation.

Thesis Committee:
Carolyn P. Rose (Chair)
Vicnent Aleven
Alexander Rusdnicky
Diane Litman (University of Pittsburgh)

Proposal Document

Keywords:
For More Information, Please Contact:

staceyy@cs.cmu.edu