CPSC 504 Guidelines for Projects

Recall that there are two options for projects – either implementation-based or paper-based. In both, we are mainly looking for creativity, novelty, innovation, thoroughness, and of course correctness. The deliverables for the two are quite different, although they share some properties.

Implementation-based projects: The deliverables include a demo, source code together with complete instructions (README file, etc.) for installing and running the code, any data sets if applicable, or pointers to data sets if they are freely downloadable. In addition, a short, 3-5 page summary of the project is expected. The summary should describe:

• What was implemented.
• Relevant prior work it’s based on.
• Other relevant work (it may not be based on, but relevant nevertheless).
• Why it was based on whatever you chose to base it on.
• What is the novelty of your work: e.g., what were the limitations of prior art, and how did you overcome them, etc. Limitations could be w.r.t. performance, functionality, or model, etc.
• What your “system” can do well and cannot do well?
• Conclusions: what have you learned from this project.
• Interesting/promising issues worthy of future pursuit.

Paper-based project: The deliverable for this option is just a paper. It’s useful to think of “standard” conference papers as a template for this purpose. Take SIGMOD/VLDB/KDD conferences as examples. The suggested length is 12 pages, two column format using SIGMOD or VLDB style/font (which can be downloaded from those conference websites). The key difference with the actual conference papers is two-fold: (a) Your work is work in progress; (b) It does not have (but is welcome!) to include an experimental component, unlike most SIGMOD/VLDB papers.

Your paper will still contain most other standard sections (e.g., Introduction, Motivation, Related Work, Background Notions (or Preliminaries), Technical Sections, Conclusions and Future Work + Bibliography. Your emphasis should be on discussing and critiquing prior art and establishing the novelty as well as advantages of whatever ideas/algorithms you propose or whatever results you establish. In other words, you need to position your work properly in the context of prior art. As mentioned above, work in progress is fine, but you should clearly identify which parts of the work are settled and are therefore solid, and which parts you are stuck in, or given more time you would have solved, what ideas you have for solving those parts, etc. Use your imagination and creativity for structuring this part.

The suggested due date for the projects is the last day of classes for this term. I am willing to extend this by a week or so. But for safety’s sake, you should plan pretending you’d get no extension.

Recall that we have scheduled “Project Talks” at the end of the course. If your project is not complete by then, you should still use the slot to present your current status of work at a high level, and use that opportunity to get a feedback on your approach from everyone. Let me know if you have any concerns or questions.