MOOC Participants who liked this post, also found this useful....

Jeeves will point you to the right discussion forum
A couple of years ago when I was putting pen to paper and I was working on my Academic Check-ins paper I was doing some more research into recommender systems, you know the systems like the ones that they have on Amazon.com and Netflix whereby if you rate a certain product in a certain way, or if you view certain products, more recommendations come up based on your usage pattern of the system.

Now, those systems aren't perfect by any stretch of the imagination, but they can serve as ways of finding some diamond in the rough that you didn't know exist.  Think about it, both in a shopping or entertainment venue, and a MOOC you have one potentially huge issue: limited time to devote, a large sea of information to go through in order to find what might entertain you, or pique your intellectual interest and  get you engaged with some subject.  Last summer, at the end of Campus Technology 2013, I was having food and drinks with new friends and colleagues that I met at this conference.  I brought up a suggestion: what if we could develop a system that could help learners cut through the noise? A system, based partly on linguistic corpus analyses of the learners or work, as well as learner psychometrics and learner and learning analytics.

The way that I articulated the system last summer would be as follows:  Learners who are participating in MOOCs, be they cMOOC, xMOOC, pMOOC, or whatever other variety comes our way, would be able to connect their twitter, blog, Google+, disqus, and facebook accounts and the system would be able to to an linguistic analysis of their posts on these services and compare them to other MOOC participants in the same MOOC who these learners should be interacting with.  This could be based on levels of educational homophily (same-ness) that learners exhibit through their posts.  Here, learners can act as More Knowledgeable Others to help each other grow as learners. To ensure that there isn't a danger of groupthink, the system would also throw in (through a magical algorithm) people with differing points of view. This way learners would have the option to read dissenting views, and hopefully engage intellectually with that aspect as well.  The level of difference could, conceivably, be something that the learner, the owner of their educational match-making profile, have control over. So if a learner feels comfortable only being stretched so much right now, they can control the level of difference that they are exposed to.

To take it one step further, learning management systems for MOOCs, such as coursera, udacity and EdX, could tie into this system. This way there is one big dashboard of learning analytics and corpus data that be analyzed to help the learner discover interesting peers, hot discussion topics and interesting topics for the learner to participate in the discussion forums of those services. So, if I am following #edcmooc for example on coursera, and Jeeves knows about it (nicknamed this system Jeeves), then Jeeves would be able to see what I am writing on my blog about this course, how I am reacting to the materials and peers on twitter, the facebook group, and how I am up-voting or down-voting some threads, and through a daily email I could be told which threads are "on fire" and that I should look at, and which threads or peers I might want to connect to, follow, or respond to. This type of adaptive system would be learning not just from one MOOC, but all MOOCs I've participated in. And, if I want to, from my blog in general.

The system would be portable, and have APIs to hook into any MOOC platform. This would ensure that the person in charge is the learner themselves, not someone else.  In the summer I was thinking of a simpler system, maybe built into something like gRSShopper or a MOOC platform, but since then I've been thinking that data portability and cross-system compatibility is much more important given the plethora of MOOC platform providers cropping up around the world.

What surprises me is that this idea is something that hasn't gotten any traction yet. I did say it in a public venue among fellow learning enthusiasts. I wonder why I haven't seen anyone else pitch this yet. Your thoughts?

Comments

It surprises me too that no-one has yet implemented a recommender system for learning objects. Most repositories have some kind of rating system but not even the slightest effort to weight the ratings in terms of the rater's closeness to one's own pattern of past ratings (let alone anything as sophisticated as what you are suggesting). I think such a system might well be of more general interest and should be implemented outside the scope of any specific platform, and I hope to hear more of your progress on this initiative.


P.S. Re the name Jeeves: Do you have any connection with ask.com (aside from the fact that apparently someone called Apostoulas(!) Gerasoulis was a co-creator of their algorithmic search technology)?
Thanks for the comment :) I am actually not related to Ask.com (formerly askjeeves) I just liked the character and tied into that past of ask.com

Popular posts from this blog

Academic Facepalm (evaluation edition)

Discussion forums in MOOCs are counter-productive...well, sort of...

Latour: Third Source of Uncertainty - Objects have agency too!