Panel Abstract and Description

Participants

Brett Shelton, Utah State University, OER Recommender
Joel Duffin, Utah State University, OER Recommender
Mimi Recker, Utah State University, Data Mining in Educational Applications
Frank Shipman, Texas A&M University, Visual Knowledge Browser/Interest Profile Manager (VKB/IPM)
Todd Will, New Jersey Institute of Technology, General Recommendation Engine (GRE)
Keith Maull, University of Colorado, Customized Learning Service for Concept Knowledge (CLICK)

Moderator : Tamara Sumner, University of Colorado

Abstract

Recommender systems provide a host of new techniques and technologies for enhancing the value and utility of digital library resources.  This panel will explore several ongoing recommender system projects within NSDL and examine future collaborations and opportunities.

Description

Recommender systems research is at the fore of next generation contextualized search technology. Within educational contexts, recommender systems offer a wide array of useful techniques for personalization of educational learning materials, and in particular the personalization of digital library resources. Within NSDL there are several on-going recommender system research projects underway, and this panel will explore the work of four such NSDL research projects : OER Recommender, General Recommendation Engine (GRE), Visual Knowledge Browser/Interest Profile Manager (VKB/IPM), Customized Learning Service for Concept Knowledge (CLICK) and Instructional Architect Web Metrics and Mining. Each of these projects has a distinct research focus and through moderated discussion, the panel will examine the technical strengths and challenges of their respective projects. Collaboration opportunities will also be dis-cussed, with particular interest on technical convergence and new uses within NSDL moving forward. We invite broad participation from other NSDL projects. Several on-going projects have been identified that can either be classified as recommender systems, or can be said to rely on techniques and tools similar to those commonly used in recommender system technology. There are two broad areas of interest for the panel : (1) recommender systems as integrated services to NSDL and (2) recommender systems as external tools that utilize NSDL resources. Within these broad areas some typical themes emerge. First, contextualization is a common aim of recommender systems, and within NSDL contextualization can take many forms. For example, OER Recommender provides links to related OpenCourseWare within existing NSDL pages. By providing links to related materials, NSDL content acts as a conduit for access to other learning-relevant resources (presumably outside NSDL's scope). On the other hand, GRE contextualizes NSDL resources based on a user profile that may changed based on the user task. A second common goal of recommender systems is personalization. IPM, for example, builds a profile of user interests based on actual session interactions with NSDL documents. This kind of personalization improves the effectiveness of NSDL resources by more accurately targeting resources to a specific user or profile. Similarly, within the CLICK system NSDL resources are selected for the end user through analysis of student writings to categorize resources that are most topic appropriate for the concepts within their writing. There is significant convergence among the varied recommender projects that will make a panel of this kind relevant. Furthermore, examining the diverse approaches each of these projects has taken should bring more interest and information to the broad potential they each have across NSDL projects. The goals of this panel can be divided into three distinct areas. The primary goal of the panel is to inform the NSDL community at large of the current recommender system research. As an outcome of the panel, a research summary of extant recommender system projects in NSDL will be made available. The second goal of the panel is to articulate the contribution, convergence and collaborative opportunities among existing and new research projects. There are wide-ranging collaborative research tracks that can be explored once a clearer understanding of the strengths of each project are identified. The final goal of the panel is to develop a proposed NSDL recommender system research roadmap. Recommender system research within NSDL is reaching a crucial point and proposing a roadmap for the future of such research will not only make NSDL's use of research outcomes more effective, but also guide future funding opportunities, research goals, objectives and collaborations.