Chair: Ken Anderson (University of Colorado at Boulder, USA)
Enhancing Access to
Open Corpus Educational Content: Learning in the Wild (Long Paper)
Seamus Lawless, Lucy
Hederman and Vincent
Lack of relevant and accessible digital content hampers the implementation of e-learning. As these eLearning tools begin to try to offer personalization, the tools require access to an increasing amount and variety of content. eLearning educators are compelled to generate their own content, which can be an excessive workload.
Trends –content creation moving from the linear authoring of publication to the aggregation of existing; rise of the prosumers, who produce an consume content in increasing volumes.
WWW already holds content useful for incorporation in eLearning options, but the issues of content discovery, repurposing, mean that even WWW content isn’t an easy solution.
Address the reliance of eLearning systems on bespoke proprietary content. Open content availability reduces the need for educators to reinvent the wheel every time they create a course. Addresses the information overload in eLearning experiences. Help students identify what is actually relevant ot them in their specific educational context.
Open Corpus Content Service — OCCS — WWW and selected digital content repositories. Discovery and harvesting of content — open-source web crawler, JTCL and Rainbow classification. Indexing with NutchWAX. Visualization — didn’t catch the acronym.
Train the Rainbow text classifier – this dictates what gets included in the cache of content.
[My humanities-trained mind is crying out for examples! I'm putting a lot of conceptual information in temporary storage caches, but the buffer is running out of room. The speaker is actually very good -- but I'm waiting for the payoff that I'm conditioned to expect a humanities presenter would have started to deliver by now. I'm learning just how important the little chart with inputs and outputs is as a convention in scholarly presentations in this genre. We're spending a lot of time on the left-most edge of a rich flowchart that I gather will start moving across the page... we're still on "Training." there we go, now we've got the "Crawling" section. Steve sitting next to me is looking up terms the speaker is using... I'm net yet sure I need to put that level of new information in my neural net until I've seen what it all adds up to.]
Okay — now we’re being walked through an example crawl —
The educator prepares the crawl by identifying the subject area, with seed file generation and training set generation Ran for almost 2 days, found 370,000 + URLs, passed some 67,000 on for further processing, judged 36,000 at 90%. Had human subject matter experts evaluate the returned content to find out how well the computer’s predictions mathed the human expert decisions.
[Drat... at this point the Seamus says he's not going into heavy detail -- yet this is exactly content I was waiting for. This is the material I'd like to have seen so that I understand what the system is designed to do, but it's what he rushed through because he judged it as less important. Steve just shut his laptop. Coincidence?]
U-CREATe interface integrates a link to OCCS.
[Okay.. I think I've finally made the phase shift. I came to this talk expecting to read a book. Instead, I got a very meticulous description of new tools for constructing books. Or, to pick a different metaphor, I came expecting to watch a dance, and I got an detailed analysis of how muscles work on the cellular level. Now we're getting usability results -- the convention of the scienctific research paper is to deliver the conclusion last, but humanities papers start with the thesis (the answer to the research question).]
Allan says his work focuses on linking physical places. How to do digital physical linking using 3D barcodes. Present programs built with this infrastructure.
Background — trying to use ubiquitous hypermedia to support urban web applications — want to let users brows and create and share information while they are mobile in an urban environment. Not just browsing, but browsing information related to the urban environment where they are.
Anchor information in the physical world; identify aspects of the physical world that we can use to anchor our links. GPS offers one sensor useful for anchoring links.
Ubiquitous link anchors: ID Mapping. Not a static model; we specify an anchor value and the system finds resources that match that anchor value. The 2D barcodes [a pattern of squares, not bars -- that name 2D barcode seems oxymoronic -- new to me, but an established term.] provide a visual anchor for the link. A URL can be converted in to the 2D barcode, scanned by a cell phone, and used to deliver a resource.
Examples… TagBlogger — 2007 Arhaus festival, lets users access official location-sensitive information; create and share digital overlays. Had to develop the software and deploy 2D codes in the city. Tags on official festival posters; also tags along a route [pedestrian, I presume].
[I wonder... did the barcodes get vandalized? At any rate, sounds like an interesting project, and far more workable than the old CueCat debacle, which would have required people to carry a specialized device around and tether it to a computer.]
Conor presents. There are physical simulations (you learn about the physical object); procedural simulations (flying an airplane); soft skills (take place in a social context, based on interpersonal relationships). Sales, interviewing, leadership.
Typically you get a short clip of the person being simulated, the learner takes on a role within the simulation, and learns by doing.
The simulations are cost-effective, convenient if online, save, educationally effective.
Demo [Thank you for giving the example this early!]
Teaching psychiatric medical students how to deal with patients (this is PARRY for the 21stC). Looks like the same mechanism for adaptive tree fiction.
Difficulty with soft-skill simulations — difficult to compose. Not only the complexity of the dialog, it also has to be educationally sound simulation. [I wonder if anyone in our family therapy program would be interested in a tool like this.. .obviously I'm interested in the ability to create a model interview for journalists, but the mechanisms will likely be similar to what a family therapist might face.]
VISIOn Composition Tool; Experience Builder; Captivate 3.
- VISIOn sems to be an outliner
- Experince Builder is accessible through a web browser.
- Captivate 3 most typical type of composition tool out there… approach is more towars hard skills, limited soft skill applications. No back links — artifact of the procedural hard-skill origins of the tool.
Conor then went through each option and evaluated each on the key requirements… the lack of backlinks makes me completely reject Captivate 3 for my own interests, and to be honest I think I’d feel more comfortable working out the structure in Inform 7.
Will present more about the ActSIM composition tool.
Mark B – the sentimental novel is intended to teach us how to act and behave in certain situations. How does an environment devoted to writing hypertext differ from an environemnt designed to teach soft skills?