Sat 12 May 2012
One of the limitations of open educational resources (OER) is that you can’t fill a curriculum with them. 3 minute YouTube videos and a pdf file don’t replace a semester of learning design and scaffolded study.
It’s long been recognised that OERs don’t make courses. Randomly combining OERs leads to overlaps and gaps which is an inefficent and time consuming way to learn. Especially in higher, more complex learning activities, away from introduction 101 courses, it is an illusion to imagine that teachers will spend time evaluating and mapping out OERs into a “learning map” that students freely choose from and which leads to the desired equitable learning outcome.
Now, a new project attempts to tackle this issue. OpenDiscoverySpace (ODS) will use linked data and analytics approaches to track the usage of OERs by learners. Methods like relationship mining and sequential pattern mining are envisaged to lead to the crowdsourcing of meaningful learning paths through the OER landscape. By identifying links between objects that users tend to follow these patterns can be detected and be recommended to future users. Say, if a users goes from OER1 on to OER2, this path can be suggested to peers.
In this way, when scaled up, as ODS tries to do, a map of OERs can be created that would take at least part of the burden from the teachers. Prediction of learning paths based on mining techniques will allow for easier recognition of valuable and suitable learning objects, and has the potential to surface invisible connections between OERs from different sources.
The strength of ODS is it’s massive scale unifying virtually all previous efforts to bring OERs and learning repositories together. This not only refers to the learning objects, but also to the learners. By accumulating and analysing large amounts of learner behaviour in OER environments and combining them with knowledge about users (e.g. what language background they have), analytics methods are hoped to reveal optimal clustering and combinations of OERs.