A user-transaction-based recommendation strategy for an educational digital library

Abstract

The automated recommendation of content resources to learners is one of the most promising functions of educational digital libraries. Underlying strategies should take the individual progress of the learner into account to provide appropriate recommendations that are meaningful to the learner. If presented with appropriate assistance, learners will more likely engage in productive learning strategies, such as reading up on concepts and accessing preparatory materials, and refrain from unproductive behavior, such as guessing on or copying of homework. In this exploratory case study, we are analyzing transactional data within an educational digital library of online physics homework problems and learning content. The sequence of events starting with a learner failing to solve a particular problem, interacting with other online resources, and then succeeding on that same problem is used to identify potentially helpful resources for future learners. It was found that these "success stories" indeed allow for providing recommendations with acceptable accuracy, which, when implemented, may lead to more productive learning paths.