Volume 5 Issue 1
Spring 2009
ISSN 1937-7266

OER Recommender:
Linking NSDL Pathways and OpenCourseWare Repositories

Joel Duffin

Models for Thought, Inc.
PO Box 529
Millville, UT 84326

Brandon Muramatsu

Massachusetts Institute of Technology
77 Massachusetts Avenue
Cambridge, MA 02139

The OER Recommender (www.oerrecommender.org) is a web service that helps people find relevant open educational resources. It links the digital learning resources in the National Science Digital Library (NSDL) disciplinary pathways with courses in OpenCourseWare repositories thereby providing critical contextual information. When a person browses a web page in a participating NSDL Pathway or OpenCourseWare repository, the recommender annotates the page with a “Recommended resources” link.

Context is critical to the use of any resource, especially the National Science Digital Library (NSDL) and its disciplinary pathways. Context can transform resources in the NSDL into teaching and learning materials. Context provides the glue that binds individual concepts and activities together into a coherent whole. The OER Recommender works to link together course materials from OpenCourseWare Repositories and individual learning resources from NSDL and its pathways to help provide context for their use by learners and teachers.

The poster describes the motivations for the project, provides detail on the recommendation engine, displays sample recommendations for participating collections, and describes how other collections can participate in the project.

For a larger view of the poster, click here. After the image opens, click on it to zoom.

The recommender uses four phases to arrive at recommendations: (1) parse metadata (acquired via OAI-PMH and RSS), (2) calculate local term weights, (3) calculate global term weights, and (4) calculate similarity scores. The recommender engine uses a standard Term Vector Model approach, which can be extended into a Latent Semantic Analysis approach at a later date if necessary. For each pair of resources that are related enough, the recommender uses the titles, description, and tags to calculate a score indicating how related they are. Recommended resources are the ones scored to be the most similar. The similarity of two resources is based on an automated analysis of the words in their metadata.

The recommender provides two methods to receive and display recommendations—sites can insert a line of JavaScript into their web pages to display recommendations and then style the XML-based recommendations for display. Or individual users can use a Firefox GreaseMonkey script to automatically display recommendations on any participating web site. Both methods are designed to provide the maximum service, while requiring a minimum of effort on the part of the user.

Acknowledgements: This material is based upon work supported by the Andrew W. Mellon Foundation, the William and Flora Hewlett Foundation and the National Science Foundation under Grant No. 0532895.