UWM Multimedia Software Lab

1
UWM Multimedia Software Lab

• Ethan Munson
• Dept. of EECS
• UW-Milwaukee

2
Laboratory Goals

• Study and create novel software to support better
  multimedia authoring and browsing tools
• Style sheets
• Software environments using hypermedia
• Multimedia document analysis
• Emphasis on language-based solutions
• Document specification languages
• Style sheet languages

3
Laboratory Funding

• Grant from NSF CAREER program
• supports one PhD, one MS student
• Subcontract from Center for Adaptive Data-Driven
  Information Processing (U. Maryland, Baltimore
  County)
• Supports one undergraduate student
• Possibly others
• Various equipment and software donations
• Sun, Macromedia, Eaton

4
Laboratory Facilities

• Laboratory space 46 m2, open format
• Computing facilities
• Sun E250 file server (2 Sparcs, 80 GB RAID,
  Solaris, AFS)
• Sun Ultra 10 (4), SGI Indigo 2 (2), HP 715 (2)
• Wintel, G3 Mac
• Digital video, camcorder, scanner, printers

5
Laboratory Entrance
6
More Pictures
7
Still More Pictures
8
Our File Server
9
Projects

• Software Concordance
• CADIP research
• Web image search
• Usability of information visualization
• User tolerance for GUI delays

10
Software Concordance

• Goal Improve software development process with
  hypermedia services
• Building Java source-code editor that supports
  embedded hyper-links and multimedia
• Integrated parsing and version control
• Links defined manually
• XML for non-source-code documents
• Will build analysis and visualization tools

11
Software Concordance

• Current status building editor using
• Fluid program analysis suite (CMU/UWM)
• Chimera hypermedia toolkit (Colorado)
• Kaomi multimedia authoring toolkit (INRIA/Opera)
  ??
• Tien Nhut Nguyen (PhD student)
• Hypermedia infrastructure
• Satish Chandra Gupta (MS student)
• Source code editor
• Interface with Fluid

12
Retrieving Web Images

• MS thesis by Yelena Tsymbalenko
• Use HTML metadata (HTML source) to identify
  images relevant to one-word text query
• no image downloading or processing required
• Initial results
• High precision for some simple forms of metadata
• Structural clues not effective
• Possible methodologic problems
• Current status need student

13
Information Visualization Usability

• Many visualization techniques for IR data have
  been proposed
• Stereoscopic Field Analyzer (SFA) is one
• requires 3D goggles and 2-hand magnetic tracking
  controls
• Goal test usability of SFA on realistic data
  sets
• These systems have high face validity, but
  little rigorous testing
• Current status beginning research now
• Nicholas Bohne (undergraduate)

14
User Perception of GUI Latency

• Goal understand how fast a GUI must be to make
  users happy
• Is as fast as possible necessary or even
  correct?
• First study instrumented MS Word to control delay
• Weak results due to bad methodology
• Subjects words dont match their survey results
• videotape may have been a problem

15
Current Latency Study

• What is minimum delay that a user can perceive
  for each GUI task (menu, button, typing)?
• Common wisdom 100ms
• Initial results 140-180ms
• Scientific issues
• Turns out to be difficult to study
• Subjects have trouble determining whether delay
  occurred
• Does perceptible delay vary with task?
• Some evidence, but methodologic problems make
  result uncertain

16
Future Latency Research

• Improve experimental technique
• Measure full probability distribution
• Long, boring testing with small number of
  subjects
• Test more GUI tasks
• to try to find task differences
• Move to studying user preference, rather than
  perception

17
Other topics

• Formal models of hypermedia and hypermedia design
• Two MS theses, possible collaboration with
  Brazilian researchers