Theme Areas:

My Prior and Current Involvement with KBEs

During my 6 years at the Learning Research and Development Center, University of Pittsburgh, and continuing into the present, I led (under the mentorship of Alan Lesgold and with the assistance of Arlene Weiner and Eva Toth) the development and classroom trials of the Belvedere software and associated materials for classroom implementation. Belvedere is a diagrammatic environment intended to support secondary school children’s learning of critical inquiry skills in the context of science (Suthers et al. 1997).

Belvedere's core functionality is a shared workspace for constructing "evidence maps," which relate data and hypotheses by evidential relations (consistency and inconsistency). The diagrams support collaboration by providing a shared context and reference point, and provide the teacher and the computer with a basis for assessing students' understanding of inquiry in general and of a topic area in particular. The evidence maps are stored on a server, and may be edited concurrently by multiple students. Belvedere was evaluated first in our laboratory and in local schools, and then used more extensively in U.S. Department of Defense Dependent Schools in Germany and Italy.

Initially, the Belvedere project focused on rich representations intended to enable these students to express scientific arguments, under the guidance of a machine coach. During our work with students, a refocus on collaborative learning led to a major change in how we viewed the role of the interface representations, and a consequent simplification of these representations. Rather than being a medium of communication or a formal record of the argumentation process, we came to view the representations as resources and guides for learners' conversations (Suthers 1995). These observations, coupled with the fact that other projects with similar goals were using radically different representational systems (compare Belvedere, CoVis, CSILE, SenseMaker, and WebCamile), led me to propose a more systematic study of the ways in which these different representational systems can influence collaborative learning discourse. This work has been funded by NSF's Learning and Intelligent Systems. Preliminary results are encouraging, and will be presented at the main conference (Suthers, 1999).

Three other projects also provide relevant background. I am author and co-PI (with Lauren Resnick and Alan Lesgold) on "NetLearn," a DOE Technology Innovation Challenge grant to develop networked distance collaboration software in support of the work of the Institute for Learning, a national consortium of (mostly major urban) school districts involved in a systemic reform effort. The work of the Institute involves rich and long-term conversations about instructional practice, these conversations being guided by various "tools" developed by the Institute, and grounded in examples of teacher and student work. The challenge of expanding this discourse to electronic media has, like the Belvedere work, led me to consider the relationship between discourse and artifacts more carefully.

The selection (if not development) of media for distance collaboration and knowledge building for university level education is a recent interest. Our department has made a commitment to develop a completely asynchronous, internet based version of its M.S. and B.A. degrees over the next few years, assisted by Sloan Foundation funding which I recently obtained for the purpose. Our special concern in this context will be learning to support highly interactive project work, not just delivery of traditional "textbook, lecture, and exam" forms of pedagogy over the wire. Such work requires sharing and discussion of multimedia artifacts pertaining to the projects.

Finally, my students and I are redesigning Belvedere's architecture, using component technology to facilitate the implementation and local customization of Belvedere-style systems that support multiple forms of models (beyond evidential models) and mutliple views of these models (beyond graphs). This architecture will enable us to more quickly adapt to unanticipated needs to share collaboratively constructed knowledge representations. I hope to take advantage of this capability in future years as part an effort to engage students in rural Hawai'i schools in the state's rich natural science and technology resources via distance collaboration and remote sensing technology. I recently obtained a Rural Systemic Initiative development grant for this work, known as Hawai'i Networked Learning Communities.

Ideas Concerning KBEs: Coordinating Discourse Media with Representational Scaffolds

The foregoing projects, though diverse in population (K-12, university, and adult professional development) and topic (earth science, computer science, and education), manifest common issues in the design of computer media for knowledge building. My focus is on the relationship between discourse, artifacts, and representations; specifically two issues:

The relationship between computer mediated discourse and artifacts under discussion: In our NetLearn work, it has become clear that standard computer mediated communication (CMC) software is entirely inadequate for artifact-centered discourse. We need better interactive and representational aids for linking on-line discourse to the artifacts that they reference and make sense of, while also not tying the discourse excessively to single artifacts, but rather enabling participants to identify patterns across multiple artifacts. This work is at the stage of needs identification: I don't yet have solutions to offer in this workshop.

The relationship between participant-constructed knowledge representations and discourse (whether computer mediated or not): In the Belvedere work and the subsequent NSF-funded studies I am concerned with the effects that representational systems might have on the focus and course of discourse between learners. Specifically, the independent variables are ways in which representations can express and make salient certain kinds of information at the expense of others. This work is presently concerned with effects on verbal discourse between co-present interlocutors. Planned work will study similar issues in textual CMC media. Along with this empirically based work, I also am interested in explorations in design of discourse media that is better coordinated with representational scaffolding. Traditional CMC (including distance learning media) is weak in this regard. It provides relatively unstructured media for synchronous and asynchronous discourse, but no attention is paid to how deliberate construction of representations of the knowledge gained by the discourse might improve learning outcomes by focusing the discourse and encouraging reflection, let alone how the representational toolkits for this construction might guide knowledge building. Furthermore, I believe that it is not enough just to provide some representational scaffolds alongside the discourse in CMC, as we did in Belvedere. Linkages between the less structured discourse medium and the highly structured knowledge representation medium might help learners move more effectively between the two. Further work is needed to know how to optimize the design of these linkages.

I am (or will be) prepared to provide examples of all of the issues I have discussed above, drawing on the Belvedere, Representational Bias, and NetLearn projects. Taking a broader view of the role of artifact and representation with respect to the workshop organizers' question concerning how KBE differ from CMC, I believe that KBE should go beyond simply providing for electronically mediated discourse. KBE should also provide for both the study of artifacts as empirical sources of constraints on ones' knowledge building and for representational scaffolding as guides to the construction of explicit expressions of one's evolving knowledge.