Magic marker
 |
| Writemarker may make cheating a thing of the past (Photograph: courtesy Kevin Phillips) |
| Research team: |
| Dr Simon Dennis and Dr Peter Kwantes (Key Centre for Human Factors and Applied Cognitive Psychology). Dr Michael Harrington from the School of English, Media Studies and Art History was part of the theory development team. |
| Funding: |
| » 2000 ARC small grant ($19,999) » 2000 UQ grant ($13,001) » 2000 Defence Science and Technology Organisation ($20,000) » 2000 portion of ARC Key Centre grant ($60,000) » 2001 ARC large grant (now called Discovery grants) ($45,000) » 2001 Defence Science and Technology Organisation ($20,000) » 2001 portion of ARC Key Centre grant ($61,500) |
| Email/Web link: |
| s.dennis@humanfactors.uq.edu.au
administrator@humanfactors
www.humanfactors.uq.edu.au |
| Faculty of Social & Behavioural Sciences |
Its creators say the technology could also help detect plagiarism, assess short-answer questions and, by measuring the timing of keystrokes, identify the person writing the essay.
Initially, the system, at prototype stage, is being proposed as a second marker. But its potential application goes well beyond that.
Its developers believe it could be used by private testing companies, state education boards, schools, universities and corporate training organisations.
They are already speaking to one state body about the possible use of the technology underpinning the system for final-year school exams.
The project is being led by Dr Simon Dennis in the Key Centre for Human Factors and Applied Cognitive Psychology based at UQ.
Dr Dennis said the big advance in the system, Writemarker, was its ability to identify factual information.
That would set it apart from systems such as the latent semantic analysis in the United States which takes words in isolation.
"It's capturing the relational information or factual information behind the sentence," he said.
"So it's looking for the facts when you're making comparisons between essays."
Essentially it looks at three elements-lexico-semantic content of the individual words, syntax and factual content.
To use the system, a lecturer would write a model essay or assignment. Student essays would be compared with the gold standard by running them through the system.
Essays would therefore have to be submitted electronically.
Apart from overcoming the obvious labour problem when marking thousands of essays, Dr Dennis believes the system would also enable much more feedback to students.
"Because of the overwhelming nature of the marking problem, it means that limited feedback is given on essays," he said.
"There's just not enough time for tutors to be able to give feedback. At the moment, there's no alternative to having someone read them. And that takes up a lot of time."
Automated essay-marking could also provide greater consistency and leave less room for marking disparities.
But Dr Dennis said Writemarker could not be used in all situations.
Its greatest application for universities would be in undergraduate subjects with large enrolments.
Essay topics would have to be fairly narrow, a more likely scenario in first-year, high-enrolment subjects.
He said infrastructure could be a hurdle, but he did not believe it was an insurmountable one.
The centre has taken out a provisional patent on the syntagmatic-paradigmatic, sentence-processing technology and is looking for commercial partners.
The overarching project, funded by the Australian Research Council and the Defence Science Technology Organisation, is to develop technologies for language comprehension. It began with work in information-retrieval and has led to spin-offs such as the essay-marker.
Dr Dennis said while it was still at prototype stage, he was confident the technology could deliver an accurate essay-marking system.
"The key insight is to say that language interpretation is actually a memory-retrieval exercise, which is not the way that it is commonly thought of," he said.
Article by Dorothy Illing, courtesy of The Australian, Wednesday February 28, 2001, page 33
