CS 5764: Information Visualization

Semester Project

The goal of the semester project is to design, develop, and evaluate a new information visualization.  The purpose is to gain experience in applying information visualization to a difficult problem and contributing novel research.  The project is expected to be a significant effort, with useful quality results. This will involve much creativity, teamwork, learning about related research, planning and implementing a solution, and writing and presenting results. All project work is in small groups. The project can easily initiate or link to your thesis research.  Good projects can result in publication.

The Topic:  

The project dataset is here:  Blue Iggy scenario.  It contains about 1500 documents.

The domain of this project is visualization for intelligence analysis.  The intelligence analysis exercise we did in class on the Stegosaurus scenario should give you a good idea about the types of problems encountered in this domain. This is a large and complex problem, so you will likely want to focus on a specific piece of the overall problem. Your goal is to create a visualization tool that will help you (and others) to solve intelligence analysis problems like this one. Your final product should be a functioning tool, and you hypothesis answer to this particular analysis problem. 

At the end of the semester, we will have a Live Contest.  At the contest, the teams will be given a new similar but smaller dataset that you will attempt to solve with your project tool during a 2 hour session.  Thus your tool should be developed such that a dataset with similar formatting can be loaded. The contest is where you find out how effective your tool really is!  The team to solve the contest dataset first will win the grand prize.

You are welcome to use any design tools and implementation environments as you see fit.  I encourage groups to make use of the VT GigaPixel Display or other advanced technologies as part of your design.  Access to facilities can be arranged.  Impress me.

The Teams:  

You will work in teams of 4 students.  Form teams during the 1st week of class.  I also encourage multiple teams to work together in a coordinated way, so that their final products can link in some way.  For example, two teams might work on two different aspects of the problem, so that the combination of the two tools creates an even more powerful solution.

Teamwork can be difficult!  It is helpful to clearly identify how each team member contributes at each stage of the project.  At the end of the semester, I will ask each team to list each member's contributions. Typically, all members of a team receive the same project scores, except for clear occasions.  Report any team problems to the instructor early, so something can be done to remedy the situation before it is too late.

The Process:  

The steps of the process and deliverables are as follows (due dates are on the class calendar):

  1. Team formation:  Form a team of 4 students during the 1st week of class.  Hand in a list of team member names, and a team name.  Then, get started quickly!
     
  2. Design concept & presentation:  Identify the problem you will tackle, review potentially related work, and create your visualization design concept.  This is the time to think big and dream up interesting solutions.  Your final design should be thorough enough that you can begin implementation and evaluation next.  The primary output goal of this step is to convey your design to the instructor and class.  The design step has 2 deliverables:

    Note about Literature Review:  Review the research and solutions that others have done that is related to your project. The goal is to identify how your work fits into the space of the current state-of-the-art.  This will require searching and 're-searching' the scientific literature.  Useful starting points are the VT Library computer science section (which has links to the ACM and IEEE digital libraries), any relevant references in papers, the class bibliography, and other people who are experts in the domain.  www.citeseer.com is helpful for tracking references. Be thorough!  You will be surprised how much similar work has been done previously.  Include pictures. As a rough guideline, you should have 10-20 references to closely related work.
     

  3. Initial implementation:  Refine your design through feedback from step 2.  Develop an initial implementation of your refined design.  The deliverable at this step includes screenshots of the initial implementation, and a short progress report that presents the refined design, implementation status, and any changes to the project scope.
     
  4. Complete implementation: This should be your fully functional demonstration.  The deliverable is screenshots and a short progress report.
     
  5. Evaluation report:  In the evaluation phase, you will now use your tools to analyze the data (Blue Iggy scenario posted above) and generate hypotheses about the plots in the data. Be sure to track your analytic process as you go, because you will need to report on it.  You should hand in a written report by filling in this answer template.  This is essentially a longitudinal insight-based self evaluation of your tools.
    Live Contest:  Your team will also compete in a Live Contest.  At the Contest (see calendar), your team will be given a new dataset (formatted identically to the Blue Iggy) and you will have 1-2 hours to attempt to solve it using your tool. All teams will be competing at the same time. At the end of the session, answers will be compared.  The last section of your Evaluation Report should describe your experience at the Contest.
     
  6. Final presentation:  During the last weeks of class, each team will give a presentation of their final product, including a scenario-based demo, and your insights/hypothesis about the Blue Iggy data.  This is your opportunity to show off what you have accomplished and impress everybody.  It may be necessary to schedule a separate demo with the instructor to adequately demonstrate the entire work.
     
  7. Final paper and archive:  Each team must produce a final paper that documents the project and results. The paper should be modeled after typical conference papers. Use the papers discussed in class as an example. Use plenty of pictures. The instructor may invite the team to submit the paper to a conference. In general, the paper should at least include:

    Final paper should be 8 pages, using this standard conference paper format or similar.

    Submit a hardcopy of the final paper, and a zip file containing all of the project materials and deliverables (code, data, presentations, papers including Final Report and step 5 Evaluation Report, etc.) -- send me a URL to download the ZIP.