HEA-STEM Conference April 12-13 2012

The conference was held at Imperial College London, in several different buildings. The parallel sessions were organised in subject strands. Our presentation was in the Maths, Stats and OR strand in the last set of papers.

A common theme in the conference was the need for maths to be set in context for each discipline. It is a problem compounded by the students' surprise in many subjecct areas on finding that their course includes maths. Speaker after speaker reported that students engage better with maths in context, and there was a lively discussion about the most effective way of supporting students: is it better to have a subject specialist teach the maths they need for their course, or should a mathematician tech the maths? The conclusion was that there should be several people contributing to this teaching, and that if the support is removed from the location where the problem was presented, the student is likely to be less embarrassed and may seek help more readily.

Our paper demonstrated the facilities in the QTI tools for contextualising questions, and also featured the first appearance of the LTI connector embedded in an institutional Moodle - the University of Glasgow's learning and Teaching Moodle instance.

LTIQuizzes update

During our presentation at the HEA STEM Conference I demonstrated LTIQuizzes through the VLE at Glasgow University. LTIQuizzes was playing on our Amazon EC2 virtual server, however to the audience it just looked like I was setting up and using a normal Moodle module. For specialist software this is a great system, because it can be part of the main VLE as far as staff and students are concerned, however safely isolated so that it doesn't put core services at risk, or require the same platform as the VLE. Moodle is a conventional LAMP stack application, while LTIQuizzes runs under Tomcat - they could run on the same server, but they use quite different technologies so it's much nicer keeping them on separate machines.

LTIQuizzes isn't really intended for production use, but it does show what is possible. For now storage is to files rather than a database, and the QTI engine is APIS, which only supports a subset of QTI. Database storage, full LTI 1.1 support as well as the LTI 1.0 extensions for VLE persistence and course membership will all be supported very soon, and the LTI section of the code can easilly be reused by other applications. (I'm also developing PHP and C# vaersions of the LTI classes.)

Announcing QTI Works

Development work during the first few months of the QTIDI project has been progressing successfully, though you would certainly be forgiven for doubting this! As well as me being notoriously bad at finding time to write blog posts, the first development iterations on QTIDI have focused on "refactoring" all of the QTI-related software components we have been building and using on recent JISC projects so that they make a much better foundation for the technical goals of this project. (Rubbish analogy time: I've taken lots of random things, laid them on a carpet, and hit them with hammers until they break into little pieces. I'm now joining these pieces up to make something nice. Hopefully. The analogy is rubbish as I can't think of any examples where I've made something nice out of random bits on a carpet. Luckily, I'm better at doing this with software, so probably shouldn't have used this analogy in the first place.)

One of the key bits of work done so far has been a redesign of JQTI, which is the Java software library for "doing" QTI stuff that Southampton developed a few years ago. I've previously blogged about why I thought this is necessary (see http://davemckain.blogspot.co.uk/2011/06/refactoring-jqti-jqti.html) and the result of this is coming together under the not-very-original name of JQTI+. This refactoring work is now almost complete, with the exception of QTI tests, which JQTI never quite implemented fully and will be revisited in a few months.

On top of JQTI+, I'm building the replacement for MathAssessEngine that will become the main technical deliverable of this project. MathAssessEngine, which was based on the original QTIEngine, is also going to be torn apart and redesigned so that it can do all of the things it now needs to do, and do them all really well.
To reflect the scope of the work we're doing, we've decided to give the replacement for MathAssessEngine a completely new name and, after a couple of months of riotously bad naming attempts, we've decided to call it QTI Works.

I will deploy regular public development snapshots of QTI Works while it takes shape over the next few months, which you will be able to find at:

http://www2.ph.ed.ac.uk/qtiworks

If you remember to wear a hard hat, you can go in and try the first development snapshot now. This showcases the brand new (and vastly improved) QTI 2.1 validation functionality in JQTI+, as well as demonstrating the newly-refactored rendering and delivery of QTI 2.1 assessment items. (You'll have to make do with a selection of sample items for the time being... the functionality that will allow you to upload, validate and try out your own items is still written on the back of some envelopes. This will turn into real code that you can play with during the next development iteration. Hopefully!)