Possible Projects for Students in the MSL
This is a list of research projects that I would be interested in
having students work on.
Right now, my ideas fall into three main areas:
The Software Concordance
One of the main projects in the MSL is the Software Concordance, which
is developing a new kind of programming environment that brings
hypermedia to software development. The goal is to improve how all
the documents developed by a large software project are maintained as
the software evolves.
Here are some current student project ideas for the Software
Concordance:
- (Project or Thesis) Develop a Java-based tool for editing UML
diagrams that is compatible with the Software Concordance prototype.
- (Project) Enhance the Software Concordance's testbed application,
which is a simple elevator simulation and its accompanying documents.
The tasks that would be involved would be:
- Enlarge and enhance the requirements, design, and bug report
documents which are in XML.
- Write XSLT code to produce HTML output from the XML documents.
- Maintain a complete version history of changes made.
- Add new feature(s) to elevator simulation application.
User Perception of GUI Latency
We are studying how people perceive and respond to delays in the
response of graphical user interfaces (GUIs). For many years,
software developers have used a rule-of-thumb that a computer should
respond within 100ms in order to seem instantaneous. We are trying to
systematically test this rule and to understand whether
"instantaneous" is really what users want. Surprisingly, there is
some evidence that users prefer a bit of delay when using computers.
Here are some of the projects that we are thinking about. Each one
could be either a capstone project or a thesis depending on the level
of independence exercised by the student.
- Obtain population data on people's abilities to
make judgments as to how fast a response to a standard key-press
and mouse-click must be for it to seem instantaneous to users.
- Measure how certain aspects of interaction influence ability
to detect delay (specifically: location, locus of attention and
possibly contrast.)
- Measure abilities to detect delay for touch-screens (both with
and without a stylus).
- Measure specific standard GUI controls to determine if people
are more or less likely to notice delay when using them rather
than simply pressing keys or mouse buttons.
This work is being coordinated by Jim Dabrowski. It requires the
patience to work with real human subjects. It will involve some
programming, but also involves experimental design and statistical
analysis of results. Contact Jim (jimd@cs.uwm.edu) or Dr. Munson if
you would like to get involved.
Otto: Web Image Search
Two previous MS students have looked at techiques for retrieving
images from the Web. We are trying to understand how to best
determine the content of images based on the text in the HTML pages
that include them. Cheng Thao's work has given us a good basis for
two projects/theses:
- Building an image search engine: Cheng has built most of
the components of an image search engine. One next step is to put
those pieces together into a package that has good performance and a
good user interface. The goal would be to quickly return a page with
links to images that match a text query. However, the way that our
approach works, this can be difficult to do quickly (in less than 5
seconds). The challenge is to balance speed and correctness.
- Analysis of search results: In Cheng's thesis, we have
initial results from analyzing 24 image search queries. These results
have given us a good initial formula for rating the relevance of
images. However, there are lots of subtleties in the data analysis
process. A student working on this project would either
- re-analyze
Cheng's data in an attempt to build a better mathematical model of
image relevance, or
- collect new image search data and relevance ratings in order to
answer new questions and improve our methodology.
These two approaches can be merged. In either case, the work calls
for an interest in data analysis and an ability to learn about
advanced statistical techniques, such as logistic regression.
Last update: 23/04/2003