|You may request notification for Orchestration Graph - another learning workflow system.|
Orchestration Graph - another learning workflow system
By: Stian Håklev
|05/04/17 07:42 AM|
I am very excited about the fact that LAMS is still living and thriving - the 3.0 update seems like a huge improvement, and all the activity in Singapore and at different institutions sounds very exciting as well.
We are working on a new system which shares much of the philosophy and ideas behind LAMS, but also has many differences. It’s also in an extremely early stage, and not ready for any kind of real usage yet. (https://github.com/chili-epfl/FROG)
However, I see lot’s of opportunities for collaboration - I would really love to discuss ideas around the technical implementation, activity plugins etc with the developers of LAMS, and around research on educational workflows, patterns/scripts, collaborative learning, curriculum design etc with researchers. More detail below. My email: firstname.lastname@example.org
I am a post-doc at EPFL in Switzerland, working with Pierre Dillenbourg on a new platform for educational workflows. I first came across LAMS last year, and was amazed at the thought and design that had gone into the tool, and the fact that I had not known about it, despite being interested in very similar things during my PhD thesis in Toronto on a collaborative EdX MOOC for teacher professional development. However, at that time I was also saddened to see that the tool looked abandoned since 2014, with no new versions of postings on the forums etc. Today, as I was visiting the community to look at the kinds of sequences people had posted, I found a message from Ernie Ghiglione stating that things were going well in Singapore, and they had released a new version, I found lot’s of activity on the bug tracker etc. Very interesting!
Let me tell you a bit about our research: Pierre wrote a book a few years ago called “Orchestration Graphs: Modeling Scalable Education”, where he proposed a new notation for collaborative learning sequences, based on activities on three planes (individual, group, whole class), and operators that generate/modify social structures, or operate on student-generated artefacts. A big part of his books was how these graphs could enable us to scale up rich collaborative scenarios, and also connect the learning analytics back to the machine-readable pedagogical intent, to gain a better understanding of how students learn, etc.
I started half a year ago on a project to make a platform that lets teachers or instructional designers design orchestration graphs in a GUI, and run sessions with students. Our idea is to have the activities and operators be completely independent plugins, and make them as easy to author as possible, enabling an ecosystem of activities and operators (algorithms for group formation, etc). Our current early prototype (open on Github) only deals with completely synchronous graphs, where all students switch between activities at the same time, but eventually I’m thinking of three different modes:
- a set of activities done by an individual or a consistent group that can be reorganized or modified by automatic systems to enable adaptive learning/intelligent tutoring systems approach (we have a student working on optimization, personalization etc)
- a synchronous mode where the students move in unison, and you can switch between different groupings, individual/group, etc. (What we have now, and what is shown in the video)
- an asynchronous mode (for homework or for MOOCs), where you have long-running activities that students complete at different times, possibly with conditions (A before B) etc.
What becomes very interesting is how to make connection/integration points between these three modes - how do you bring individuals from an adaptive learning mode, into collaboration, if they have progressed to very different points, etc. Lot’s of nice research topics. We have two experiments planned this fall - one an integration within an EdX MOOC, where students would also author their own sequences, and one in a large undergraduate lecture (400 students) on intro to statistics, probably using smartphones.
On the technology side, I have also been in discussion with some other learning research groups that are interested in formalizing a richer way of integrating remote activities than are currently available with LTI, but without having to buy into our specific theory of graphs etc. The idea is that activities should be able to accept input in (for example a concept map can take a list of activities), live stream learning analytics (for dashboards/monitoring), accept orchestration (reorganization of students, pausing etc), and return the student artefacts generated (the concept maps etc).
I am attaching a link to a very brief demo of our system which we recently showed at a conference, together with the paper and poster from that conference which shares some details about our current design.
I'd love to hear from anyone who would like to discuss this further.
Posted by Stian Håklev
Back to LAMS for Tech-Heads - General Forum