CS 785: Course Project: Proposal

This course includes a semester-long research project, which is cumulatively worth 50% of each students grade in the class. Students are permitted to work alone or in teams of up to 3 (I recommend working in a team, because although I will have higher expectations for team projects, students usually report that working together is easier). The main goal of the course project should be to do a (small) research project that is related to one of the themes or topics of the class. I recommend strongly that students draw on their own research interests when choosing a project: I do not mind if you "double-count" the research you were going to do anyway as (part of) your course project, too, as long as it is related to program analysis. While I do expect some novelty in your course projects, the standard is much lower than it would be for e.g., a conference paper.

Form a team, select a research project related to program analysis, and write a 2-3 page project proposal using the ACM master article template for LaTeX. Your proposal should:

• Motivate the problem you are trying to solve: explain why it is important.
• Clearly state the scientific question you plan to answer or the contribution that you plan to make.
• Give an overview of your technical approach. While you don't need to have figured out all of the technical details yet, you need to convey your overall plan.
• Explain how you plan to evaluate your solution. How will you know whether you have succeeded? What experiments will you run? What metrics will you use?
• A brief survey of related work that you are aware of. I don't expect this to be comprehensive, especially if your research area is not in program analysis. However, I should be able to tell from this section how much you do already know about the area in which you're planning to work.

Notice that these are the same elements that you would need if you were proposing to write a conference paper. That's by design, and you should think of the course project as a "mini" conference paper; a very successful course project can often turn into a real one.

The possible choices for project topics are intentionally broad: I will permit just about any project so long as you convince me that it is 1) worthwhile and 2) related to program analysis. However, if you're not sure where to start, here are some general ideas about acceptable project topics:

• Proposing and evaluating a new program analysis.
• Applying a technique from program analysis to a new domain (e.g., your research area).
• Applying a technique from another research area (e.g., yours) to a problem that is important in program analysis.
• Evaluating existing program analysis tools, via case studies or via controlled experiments.
• Implementing a technique that has been proposed, but never actually built.
• Porting a program analysis tool to a new programming language.
• Extending an existing program analysis tool to better solve whatever problem it targets (hint: check the future work sections of papers describing existing tools for ideas!).

Your (and your fellow students') answers to \HW1 are a good place to start looking for project ideas. After HW1 is submitted, the instructor will post a link to all HW1 submissions here. You might also draw ideas from your own on-going research projects or from the future work sections of papers you have read, or from any other source. Don't be overwhelmed by the scope of the project: the instructor will help you in choosing a good project and in making sure that it is achievable in a semester.