READ THIS FIRST
I will be holding a briefing for all students who are interested in undertaking any of my projects:
On: Monday 1 March (Week 8)
At: 1715-1815
In: CS103 (Seminar Room)
If you are interested in any of these projects, please attend the briefing. This will be your only opportunity to discuss these projects, unless you have a good reason why you cannot attend. If you cannot attend - because of a timetable clash or because you are on a sandwich placement, then please email or phone (01904 432775) me to express your interest. You can also view the slides for that briefing.
Following that meeting, students who have definitely decided that they wish to do one of my projects will be asked to email me. If more than one student is then interested in any one of my projects, I will invoke the tie-breaker. Details of the tie-breaker mechanism will be given at the briefing.
As should be apparent from the list below, I am mainly interested in supervising projects related to human-computer interaction. If you have an idea of your own in that area that you think I might be interested in supervising, then please talk to me. I am particularly interested in:
The Royal Anthropological Institution (RAI) is in the process of digitizing a classic set of documents. (A sample is shown here). A problem with any such exercise is to ensure that the digital form will remain accessible throughout all changes and developments in technology over time. There have been a number of instances where this has not happened. A notable example is the Blue Peter programme's attempt to create a Doomsday Book in 2000. The 'book' was created on a laser disc in a format which rapidly became obsolete and there are no players capable of playing it.
The storage medium is one vulnerability. Another is the data format. At present the documents are being stored in tiff format, but what if in 10 years no one uses tiff and there is no software that can interpret it?
The objective of this project will be to find a format which is as robust as possible against such evolution. Something based on XML (Harold, 2002; Holzner, 2001) seems a likely candidate, but the exact format and representation of the images would have to be worked out. Of course it will also have to be possible to translate from tiff into the chosen format.
Outline of the project:
The ultimate success of this project will not be measurable in its lifetime. That is to say, that only in 10 or 20 years when the RAI's digital collections are - or are not - still accessible will it be apparent as to whether the project worked. It would be good for the report to include an estimate of the likely success.
A research proposal based on this was submitted to the Arts and Humanities Research Council. The bid was unsuccessful, but the Case for Support will provide a basis for this project (although, of course, the student project will be much less ambitious).
Harold, E. R. & Means, W. S (2002) XML in a nutshell : a desktop quick reference, Cambridge : O'Reilly
Holzner, S. (2001) Inside XML Indianapolis, Ind. : New Riders.
Royal Anthropological Institution (RAI) digitisation projects
Sugimoto, Go (2002) XML application for archaeological image database : an example of Herefordshire SMR aerial photographs University of York MSc Thesis.
The Advertiser Group is a local company which plans to launch a new website for the buying and selling of trucks and related goods and services (TruckAdvertiser). They have some ideas that they would like to work through with a student as to how the site could be designed to maximize selling. It is envisaged that this project will proceed in parallel with a paid web designer who will be implementing the conventional parts of the website.
At the same time, they would like to facilitate low-cost advertising on the site for small businesses. That is to say that registered businesses could create their own content and mount it on the site without the (expensive) intervention of a web designer. Of course, to be effective, any such pages must not look amateur. This aspect of the project would therefore look at the feasibility of producing a tool for webpage generation with the following properties:
The tool would presumably be form-based. It might use techniques such as skins and templates.
The project might draw on a number of areas of related work. There has been interest for some time in the concept of 'walk-up-and-use' web interfaces for casual web users (Lewis et al. 1990), but in this case the aim would be to develop a similar type of interface but for web users creating web pages. At the same time there has long been an interest in empowering untrained people as programmers (e.g. Pane & Myers, 1996; Pane et al., 2002), although web design is not a form of programming.
There will be two aspects to evaluation of the project: the usability of the authoring tool and the quality of the pages they can generate. The design of the evaluations will be an important aspect of the project and guidance may be found in Cairns & Cox (2008) and Sharp et al (2007).
The student would have to liaise with the Advertiser Group with regards to their requirements and any website/tool produced must be evaluated. It is likely that the project will require the use of CSS, JavaScript, PHP, mySQL and possibly XML.
The project might be linked to alistair.05.
Beaird, J (2007), The Principles of Beautiful Web Design, Colingwood: Sitepoint.
Lewis, C., Polson, P., Wharton, C. & Rieman, J. (1990) Testing a walkthrough methodology for theory-based design of walk-up-and-use interfaces CHI '90: Proceedings of the SIGCHI conference on Human factors in computing systems, Seattle, Washington, United States, pp.235-242
Pane, J. F & Myers, B. A (1996) Usability issues in the design of novice programming systems, School of computer Science, Carnegie Mellon University, Technical report CMU-CS-96-132
Pane, J. F., Myers, B. A., Miller, L. B. (2002) Using HCI Techniques to Design a More Usable Programming System IEEE 2002 Symposium on Human Centric Computing Languages and Environments (HCC'02), p. 198.
CSS Zen Garden is an example of a 'skinning' approach implemented using CSS. The same website can be viewed in a vast number of different styles. The underlying HTML is identical, and the different appearances are implemented through different stylesheets.
Colour can be used to convey information in a culturally-dependent way. That means that a website constructed using one colour scheme might be more effective with members of one community than another. The objective of this project would be to investigate whether this is true.
The project would consist of
The literature on cultural significance of colour is probably broad but found outside the usual Computer Science references (e.g. Gage, 1993). A substantial literature review would thus provide an important starting point. A major contribution of the student would be to apply that knowledge in the web context (as in Barber & Badre).
The project will call for a small amount of web design and programming (a little bit beyond that taught in WED since there will have to be a connection to a database), but the main emphasis will be on the evaluation. Given that the project is about the web, it has the advantage that it can be carried out on the web. That way it should be possible to get sufficient numbers of participants to get some useful results.
Barber, W. and Badre, A. Culturability: The Merging of Culture and Usability
Beaird, J (2007), The Principles of Beautiful Web Design, Colingwood: Sitepoint.
Gage, J. (1993). Colour and culture: practice and meaning from antiquity to abstraction. London, Thames and Hudson.(L 1.1 GAG)
Vanka, S. and D. Klein (1995). Colortool: An information tool for cross cultural design. Proceedings of the 39th Human Factors and Ergonomics Society Annual Meeting pp. 341-345.
Haptic or force-feedback devices, such as the Phantom Omni, allow people to create invisible objects which can be felt through the device. Such objects have haptic properties - the way that they feel when (effectively) probed with a pen-like actuator. In creating such virtual objects how is the design/programmer to know what they should feel like, both in terms of their surface texture and their deformability?
Users often resort to fruit or vegetable similes (e.g. 'like an orange'). The evidence for this is anecdotal, though. Part of the initial literature review would be to see whether it is documented
There are a number of problems, including
This project will address some of these questions (to be chosen in advance). The student would investigate the relevant properties of fruits and vegetables and then translate them into virtual models for the Omni. Then they would carry out experiments to assess people's perceptions of these virtual objects. Ultimately it would be hoped that a consistent vocabulary could be agreed on as well as tools or guidelines on the conversion from these descriptions into virtual objects.
This project will require good programming skills - to get to grips with the Omni API, but will also require an ability to design and execute experiments with human participants. A knowledge of basic mechanics will probably also be an asset.
Oakley, I., A. Adams, et al. (2002). Guidelines for the design of haptic widgets. People and Computers XVI - Memorable Yet Invisible: Proceedings of HCI 2002, London, Springer, pp. 195-211.
The Web has evolved to become increasingly interactive. Whereas it was once a source of information, which users consumed, increasingly users also generate content: in blogs, wikis, social networking sites and the like. This is one of many examples of ways in which the Web has outgrown its original conception in ways that no one interpreted. A consequent problem is that the underlying technology, based on HTML, was only ever intended to be used by people who were comfortable with the concept and realities of markup languages, but now 'everyone' is a web designer.
There have been attempts to make the notation more accessible, more easy to use. For instance, many wikis and blogs use BBCode, which is essentially a simplified, disguised form of HTML. There are some problems with it, though. From the user point-of-view it can on the one hand still be difficult to use, while on the other it may seem to lack power: the user is unable to generate the design they want, a design that they know must be possible. From a technical point-of-view the HTML generated may be of very poor quality - or even wrong.
An alternative approach is more of a mash-up, whereby users copy and paste bits of HTML, which they may not understand, but which (sometimes, usually?) generates the effect or appearance that they hope for. Again, the HTML thus generated is not likely to be of the best quality.
In both cases, particular problems can occur when text is deleted - or any edit is undone. It can be difficult for a program to calculate what is to be removed from the markup. For instance, suppose there is some text represented with the HTML below, but in a manner that the tags are invisible to the user - he or she sees only the effect of the markup.
<p><b>Some text</b></p>
Then if the user deletes the words Some text what should the editor remove from the file: just those words; those words and the <b></b> tags or the whole paragraph, including the <p></p> tags?
The first objective of this project will be to look at the usability of these approaches. Studies will be carried out with users to find how easy or hard they find it to generate pages to their satisfaction. At the same time the HTML they generate will be evaluated. Having found the limitations of the current approaches, the student will go on to investigate whether there is a better approach possible. In particular, the question is whether it would be possible to devise an intermediate code (as is often the case in programming language translation) between (something like) BBCode and HTML - and in so doing possibly overcome some of the problems outlined above.
For instance, the intermediate language should have a logical representation of the above paragraph, such that the structure of the text is clear and consistent (and hence matches the user's mental model), so that deleting the text causes a re-structuring, and that new structure will be reflected in the HTML which is regenerated after the edit. In other words, it is not a matter of the editor searching around and deleting tags in the HTML.
This project would suit a student who is comfortable with compilers (so, an undergraduate who has taken LSA and is taking or has taken CGO). Since it has an HCI component, it would also be suitable for a student on the MSc HCIT or MSc IT programmes - but only if they have some knowledge of compilers.
I can find no prior references to this kind of study, so I would suggest that the kinds of techniques advocated in Sharp et al (2007) and Cairns & Cox (2008) should be applied to this area.
BBCode Guide http://www.phpbb.com/community/faq.php?mode=bbcode (accessed 27 January 2009)
The Cervical Cell Sonification Project aimed to improve the screening of cervical cell samples (Edwards et al, 2008). Cells are screened by cytologists examining cell samples under a microscope - as in this example. Most cells are normal, but some will show signs of abnormality - which may be a sign of potential cancer. The idea behind the project was that if the cytologist was hearing a representation of the cells at the same time as looking at them, then their screening accuracy might be improved.
A question is as to what sounds should be used; the objective of this project would be to experiment with different kinds of sounds. Since most cells are normal, there should be an ambient, unobtrusive sound to represent them, but then cells can have different degrees of 'badness' depending on how likely they are to be pre-cancerous. Sounds need to be found which evoke the appropriate degree of badness.
An on-line experiment is already being attempted using one particular set of sounds, but the objective of this project would be to experiment with different sets and kinds of sounds. The existing set are quite synthetic, but it would also be possible to use ecological sounds. For instance, bird song might be used. Normal cells might be represented by a pleasant morning chorus, but bad cells by alarm calls.
It is likely that the best way of running such an experiment will be on the Web, so web programming experience would be useful, as would some knowledge of sound, sound manipulation and programming.
Butterfield, T. (2005). Improving the detection of cancer by using sonification to supplement visual displays. University of York, Department of Electronics, Fourth-year student project.
Lee, P. (2006). Image (pre)-processing in detecting cervical neoplasia (cervical cancer). University of York, Department of Electronics, Fourth-year student project report.
Podvoiskis, A. (2004). Improving the efficiency of cervical cell sample analysis using sonification. University of York, Department of Electronics, Fourth-year student project report.
Roseblade, R. (2006). Sound and interface design for improving detection of cervical cancer. Master's Thesis, University of York, Department of Electronics.
Stammers, J. (2006). Developing synthesis techniques for the sonification of precancerous cells. University of York, Department of Electronics, Fourth-year student project report.
Graffiti is a script that was designed to facilitate pen-based input on devices such as the 3-Com Palm Pilot. A requirement was that it must be possible to make every letter with a single stroke (i.e. without lifting the pen and putting it down within the formation of a single letter). It was also important that it should be easy to learn and therefore the shapes of the Graffiti strokes resemble the printed characters (in practice all capital letters except for 'h' and 'q'; see Figure 1). Finally, the whole motivation for the invention of Graffiti was that natural handwriting was too hard to recognize (with the limited processing power of a Palm) and that the letters of the alternative script should be easily distinguishable.
Figure 1: Graffiti strokes for the letters of the alphabet.
Graffiti is intended purely as an expressive script. The user never sees the shape of the Graffiti strokes except when learning the script, using diagrams such as that in Figure 1. This project, though, will investigate how well people can read Graffiti - both visually and through a haptic device.
Figure 2: A braille cell - which consists of up to six dots.
Another written script that shares some properties with Graffiti is Moon. Moon is a tactile script meant as an alternative to braille, for blind readers. Whereas braille is based on patterns of dots (Figure 2), Moon consists of lines and curves (Figure 3). The idea behind Moon is that it should be easier to learn for people who have gone blind later in life. In other words, they will have had visual experience of seeing letters and thus the Moon letters are meant to resemble the letters - rather as Graffiti does.
Figure 3: Moon tactile writing.
The availability of force-feedback devices, such as the Phantom Omni opens the possibility of creation of virtual haptic diagrams. A component that might be missing, though, is that of textual labels. Hardwick (2002) has experimented with the rendering of Moon on a Phantom and the idea behind this project is to attempt a similar implementation, but using Graffiti, rather than Moon. This is a pragmatic difference, because there are many more Graffiti users than Moon readers. (There are approximately 5,000 Moon readers, though there are probably no Graffiti users who are blind).
The objective of this project is to ascertain whether Graffiti might be used as an alternative form of non-visual writing, when delivered through a haptic device. If this proves feasible, then this might be a means of providing textual information within a haptic model. For instance, an Omni might be used as a means of presenting a street map (Moustakas et al, 2007) but this could be supplemented with the names of the streets in Graffiti.
Existing non-visual forms are braille and Moon. Braille does not lend itself to haptic rendering and Moon is only read by a few people. Given that Graffiti letters resemble the alphabet it might be easier to learn for those who have sight or who have had sight.
The first phase of the project would be to measure the visual readability of Graffiti. This will have to be established as a baseline for the future experiments. Then software will have to be developed (in C++) to present a virtual haptic Graffiti. Readability of this will be tested - preferably with some people who are skilled Graffiti writers as well as some who are not. Experiments with blind users would be valuable too, but probably not practical.
The project should conclude as to whether it is practical to use Graffiti in this way, and what dimensions of haptic Graffiti letters are most legible.
Hardwick, A J (2002) Rendering of moon text on simulated tactile diagrams for blind computer users by force-feedback (in) Touch Blindness and Neuroscience. S. Ballesteros and M. A. Heller (Eds). Madrid, Spain, UNED.
Moustakas, K, Nikolakis, G., Kostopoulos, K., Tzovaras,D and Strintzis, M. G. (2007) Haptic Rendering of Visual Data for the Visually Impaired, IEEE Multimedia.
The idea of sound graphs as a means of representing line graphs to blind people is not new, and was first presented by Mansur and Blattner (1985). The idea is normally as below, whereby the height of the curve on the y-axis is represented by the pitch of a sound, while the x-axis is represented by time. The Audio Graphing Calculator is a commercial (but free) product which implements this idea.
Evaluation of soundgraphs has been largely informal, but that is enough to suggest that they are of some utility. However, there is a suggestion that they could be a lot more useful in the teaching of mathematics, particularly to blind students. The idea is to give students a more direct engagement with the mathematics. The basis of this is that the method of interacting with the graph should be through joysticks. The suggestion is that the scope for teaching a range of mathematics is possible if reformulated to be based on line graph representations. More details of these ideas can be found a separate document (PDF).
The objective of this project will be to implement joystick-controlled soundgraphs. Realistically, the implementation will have to be a prototype with limited functionality. The emphasis will be on the interface, the utility of the joystick-based interaction.
The student would be expected to produce some sample lessons to teach some simple mathematical concepts based on soundgraph representations and to test these with users. For a feasibility study such as this, the participants will probably be capable mathematicians, so the evaluation would concentrate on the usability of the system, rather than learning outcomes. That would be left for follow-on studies (assuming that the feasibility has been demonstrated.)
The project involve programming, in a language such as C++, as well as some evaluation with users.
Bonebright, T. L., M. A. Nees, et al. (2001). Testing the effectiveness of sonified graphs for education: A programmatic research project. Proceedings of the 7th International Conference on Auditory Display (ICAD 2001), Espoo, Finland, Helsinki University of Technology.
Dobbs, J. (1991). Sound Graphs for Blind People, University of York, Department of Computer Science, Final year project report.
Edwards, A. D. N., H. McCartney and F. Fogarolo (2006). Lambda: A multimodal approach to making mathematics accessible to blind students Proceedings of Assets 2006, Portland, Oregon, ACM, pp.48-54.
Edwards, A. D. N. and Stevens, R. D. (1993). Mathematical representations: Graphs, curves and formulas. in D. Burger and J.-C. Sperandio (ed.) Non-Visual Human-Computer Interactions: Prospects for the visually handicapped. Paris: John Libbey Eurotext. pp. 181-194.
Mansur, D. L., M. Blattner, et al. (1985). Sound-graphs: a numerical data analysis method for the blind. Journal of Medical Systems, 9: pp.163-174.
Wei Wang (2009) developed a web database of tactile image to aid in retrieval and re-use of images. The database relies on hierarchical categorization of images, according to categories chosen by Wang. There are some problems with the available categories:
The American Printing House for the Blind (APH) has a similar - and more extensive - database, and has taken a similar categorical approach. In this case there is a large number of categories, more than required for the current size of the database, so that many categories are empty.
The objective of this project would be to see whether a better match might be achieved if users of the database could assign their own categories - a so-called folksonomy (Fu, 2009; Halpin et al, 2007; Peters, 2005, Pink, 2005; Vander Wall). That is to say that users could put new diagrams into the database. If appropriate categories exist, they would use them, but if they consider that theirs does not fit they can create their own. Over time it would be possible to see if a clear, coherent and usable structure emerges - or chaos, with no one able to find diagrams that they need, even when they are available.
The student would be expected to implement an on-line database (probably using PHP and mySQL). This might be based on Wang's or designed and built from scratch. But, whereas Wang's database has fixed categories, this database will allow users to select and assign their own categories. This might be based on the use of tagging (Getting, 2007). In that case, it would be possible either to allow all users of the site would to tag diagrams, or to constrain this to the diagrams' originators. In other words, the database might be something like a tactile version of flickr.
The student would implement a database with this facility and then test it. Within the scope of a student project, the testing would have to be constrained. A small number of diagrams would be provided to a small number of testers. They would insert them into the database, classifying them as they see fit. Then another set of testers would be tasked with finding particular diagrams in the database. It will be possible to collect both objective data on the efficiency of the database (is the resultant structure logical and not excessively redundant) as well as participants' subjective reactions as to how convenient it is to insert and find database entries.
Fu, W-T (2009), A Semantic Imitation Model of Social Tagging, Proceedings of the IEEE conference on Social Computing: 66-72
Getting, Brian (2007), What Are “Tags” And What Is “Tagging?, Practical eCommerce.
Halpin, H., Robu, V. & Shepherd, H. The Complex Dynamics of Collaborative Tagging, Proceedings of the 16th International Conference on the World Wide Web (WWW'07), Banff, Canada, pp. 211-220, ACM Press, 2007.
Peters, I (2009). Folksonomies. Indexing and Retrieval in Web 2.0.Berlin: De Gruyter Saur
Pink, D. H. (December 11, 2005). Folksonomy. New York Times
Vander Wal, T. Folksonomy Coinage and Definition
Wang W (2009) Development of an on-line database of tactile graphics, Department of Computer Science, University of York, MSc (IT) project.
Conventionally computers use a very limited context to support users' actions. For instance, when the command to open a file is issued, the system is likely to present the contents of the last-used folder. However, people generally use a much richer context. Indeed, folders may be the last reference that they have in mind, rather thinking about the last time they used a document, or of a document which is semantically related to the one they are currently working on. The objective of this project will be to see if it is possible to use meta-data associated with documents to create a more usable system.
This could be quite ambitious as a student project, so the plans is to rapidly build a prototype system, based on GNOME technologies (probably a document editor) and then to concentrate on a longitudinal evaluation of the system. The evaluation methods to be employed will have to be decided within the project. It would seem appropriate to use a small number of evaluators but for them to learn and use the system over a period of time. Data will be available from system logs, but it will also be important to collect usability data. This might take the form of think-aloud protocols, interviews and/or questionnaires.
Further details are provided in Felipe's specification (PDF).
Suitable as a Computer Science project.
Suitable as an MEng fourth-year project.
Suitable as an MMath fourth-year project.
Suitable for MSc (HCIT) students - usually involves HCI / evaluation work.
Suitable for MSc (IT) students.
Suitable for MSc Computation students.
Suitable for students on the MSc in Safety-Critical Software Engineering programme.
Suitable for MSc (SIIT) students
Alistair Edwards (email: alistair)
13 January 2010