Artificial Intelligence Research Methodologies
Lecturer's Self Review 2000-01

Alan Bundy


The AIRM module ran for the first time this year, although a lot of its structure was based on that used successfully in the "Advanced Topics" modules (AAR, ATCM, ATNLP), which it replaced.

Student Numbers

30 students took the module, consisting of 17 UGs, 4 MScs and 9 PhDs. The module was compulsory for students taking Linguistics & AI (3) and AI & Psychology (7 AI&Psy, 3 AI with Psy) degrees. The number of MScs was a little disappointing, especially if it is a guide to how many AI MScs are considering going on to a PhD, but maybe will improve with more advance publicity next year.

All students attended the 8 plenary lectures in the first 4 weeks. For the subsequent 6 weeks students were divided into 3, roughly equal (9--11 students), presentation groups, loosely based on research area. All groups had to run into the 10th week of term either due to timetabling problems or because of the size of the group. These were led by 4 group conveners; one group being shared between 2 conveners.

Tutorials ran from the 2nd to the 9th week. Students were divided into 4, roughly equal (7--9 students), tutorial groups, with mixed research interests.

How the Module Went

On the whole the module went well and seemed to be well received by the students, but there were some teething problems which need to be addressed in future years.

Tutors reported mixed success. In some tutorials attendence was patchy and preparation poor, whereas in others it was good. Depending on the topic of the tutorial, tutors were able to provide better advice in some weeks than in others. Because the tutorials were composed of students with a wide range of research interests there were difficulties for both the tutees and the tutors in understanding the technical background to the contributions of some tutees. This is likely to get worse as the remit of AIRM evolves to cover more of informatics, so must be addressed. Discussion in tutorials was sometimes a bit vague and abstract and there was an element of repetition between tutorials.

Presentations were generally of a high quality. The main problem reported was the logistics of getting the presentation timetable sorted out early enough to allow students adequate preparation time. The reviews of advanced papers were also generally good.

The literature surveys were also of a generally high quality with students investing a lot of energy.

The lectures seem to be well received and informed the subsequent practical exercises. The use of guest lectures worked well.

Plans for Next Year

As previously agreed this module should evolve in time into an Informatics Research Methodology module. The next step in this process is the integration of more material on Cognitive Science. As I will be on sabbatical during terms 1 & 2 of 2001/02 another lecturer will need to be found. Ideally, this should be someone from a Cognitive Science background who can develop this aspect of the module. This raises the question of what the extended module should be called -- maybe Artificial Intelligence & Cognitive Science Research Methodologies. Possibly there is a brave soul who is prepared to take this all the way to an Informatics Research Methodology module in one step, which would be wonderful, but unlikely. I am prepared to have a go at this on my return from sabbatical in 2002/03 and will give some thought to it and hold some discussions with CS folk in such anticipation.

The part of the module needing most attention is the tutorials. The tutors for next year should be identified early and some training provided before the module starts. There might also be meetings with the tutors during the term to detect problems and provide extra guidance.

I think it is valuable for the students to experience a wide range of research areas and methodological approaches in the tutorials, especially since the presentations are necessarily more narrowly focussed. Despite the communication problems this diversity causes I am inclined to persist with it.

The tutorial topics needed to be reviewed by the new module team to try to remove some of the repetition, to provide more support for preparation and to overcoming the communication problems between students from different research areas. The tutors have provided some detail feedback here, which I will pass to my successor. Better attendence and preparation needs to be encouraged. Tutor training might help here. We should consider requiring tutees to submit some record of preparation, although this would probably not be examinable and would involve the tutors in extra work to provide feedback.

Timetabling of tutorials and presentations is an inherent problem with this structure. Until one knows how many students will take the module and what their research interests are, it is impossible to finalise. I don't see any easy solution to this although the experience of this first year will give some basis for estimates of the composition of the class and allow some planning.

I would like to express my thanks to the 3 group conveners, 3 tutors and 3 guest lecturers who helped to make this module a success.

Home : Teaching : Courses : Irm : Annual-reports 

Informatics Forum, 10 Crichton Street, Edinburgh, EH8 9AB, Scotland, UK
Tel: +44 131 651 5661, Fax: +44 131 651 1426, E-mail:
Please contact our webadmin with any comments or corrections. Logging and Cookies
Unless explicitly stated otherwise, all material is copyright © The University of Edinburgh