NetQuest is a project enabling tutors and students to create sets of questions for student self assessment. The project aims to provide a large questionset which will act as a central resource, accessible over networks including the World Wide Web, marked up using TML (Tutorial Mark-up Language) so that the user can request a set of questions on any given topic.
Initial development work has been made possible by grants from the Teaching & Learning: Excellence & Innovation and Continuing Professional Development Funds of the University of Bristol and the charity Baby Lifeline. NetQuest plans to compile indexed and searchable "questionbanks" and is producing question sets in the subject areas (initially) of geoscience, chemistry, medicine, veterinary science and engineering.
TML is an interchange format designed to separate the semantic content of a question from its screen layout or formatting. The language supports several different types of question within the same content model and is essentially a super-set of HTML. Various tools allow the questions contained within TML files to be delivered to students via the World Wide Web and for responses to be tracked and analysed. Fuller details and example question sets are available from: http://www.ilrt.bris.ac.uk/netquest/
The interchange format is based on work initially done by Neil Holtz from the University of Carleton. It has, however, been extensively re-written and extended, principally via the addition of scoring and hot-image type questions. The core of this work was undertaken by Joel Crisp whilst working at the Institute for Learning and Research Technology. We are indebted to Neil Holtz for releasing his original work into the public domain. In the spirit of the best traditions of the Internet, TML remains freely available.
During the development of this interchange format, a program was written to convert the information in a TML file to raw HTML which could be delivered to a standard WWW browser. This program is a CGI-compliant program currently working with the Unix NCSA Web server; ports to both Windows/NT and Mac systems are in progress.
A TML file is a plain text file containing only HTML and TML. A number of "tags" are defined which act as semantic descriptors. These tags allow the definition of several different parts of a question, including scoring, and optionally the defaults for any section.
So far, four different types of question been implemented (Multiple-Choice, Poly-Choice, Word-Match and Hot-Image), and from the experience with these a much wider range should be possible. A typical TML file has a number of sections which comprise: a title for the tutorial, a section describing defaults, the actual question text in HTML, the correct and incorrect answers, a set of responses tied to the answers, a set of scores tied to the answers and a set of hints.
The limited number of sections provide a simple format for the file but are capable of representing a wide range of question types. Any standard HTML 3 mark-up may be used for the question type, answers, responses and hints; links to other documents, images and equations (when browsers eventually support them) may be included.
To illustrate the results of some of the development work in economics see these demonstration questions set up for Biz/ed. Anyone who is willing to contribute questions (and answers!) for the project should contact Dan Brickley at the ILRT.
The program which has been used to test the TML format is a Perl script which runs on the WWW server to convert the TML file into sequences of normal HTML pages. It takes the information in the tutorial, the feedback from the student and the saved state for the real (in "authenticated" mode) or fake (in "anonymous" mode) user ID and constructs a suitable response page. The flexibility of the program is extended by its modular architecture. Each question type is implemented in a small module, and a core set of functions exist in an `engine' file which handles logging, navigation and TML parsing.
A reporting program can analyse the results from the tests. At the time of writing, this program reports the number of questions attempted, the total score and the number of hints for each ID. It is being extended to report also on a per question basis, which can help to identify areas of knowledge which are lacking across all the students, in addition to questions which are inappropriate or incorrect.
When the tutorial engine is used in authenticated mode, an auxiliary program to enable lecturers to quickly and easily set up student passwords and classroom groups is being developed. This program uses a WWW-based front end.
Presently TML files are constructed using any editor. We are investigating other tools for authoring (e.g. wizards in Word/Excel/Access, a forms- based Web tool).
A JAVA applet is being developed to allow hot-image type questions to be authored quickly and easily.
We wish to provide searchable catalogues of questions to enable "intelligent" selection from a database.
TML is in fact an SGML DTD. The development of XML offers the prospect of presenting (and processing) TML as a client-side XML application (without having to use CGI scripts on a server to render tutorials).
We would like to provide filters to allow the easy exchange of question content from one assessment delivery system to another via the medium of the TML interchange format.
We are presently documenting a port of the TML system to Windows/NT and will then consider doing the same for MacOS. Other sites have already completed ports of TML to both these operating systems.
TML and NetQuest are open, easily modifiable, networkable and - by virtue of being Web-based - independent of platform. They offer the potential to:
TML and the support utilities are being used within a growing number of departments at the University of Bristol to delivery self-assesment materials via the WWW and it is anticipated that this form of question delivery will become very significant over the next few years.
The hardest thing about TML is constructing good quality questions to deliver with it!
Jane Williams, Dan Brickley and Helene Missou are members of the Institute for Learning Research at the University of Bristol; Paul Browning is in the Department of Geology, University of Bristol.
The authors may be contacted at
Telephone +44 (0)117 928 7096
Fax: +44 (0)117 928 7112
E-mail: j.williams@bristol.ac.uk
URL: http://www.ilrt.bris.ac.uk/