CSE 490E1 Assignment 5: Revised Project Proposal

Due: Friday, January 20, 16:59 via Canvas

The goal of this assignment is to finalize your project proposal. You will learn from and improve on the first iteration. You will ensure that your project is solving a good problem: it matters in the real world, isn't too hard, and isn't too easy. You will avoid nasty surprises later: your project will be more successful if you think through the high-level issues before you get mired in the details.

Before you start, create a version control repository for your team. A GitHub repository is a good choice. If you want something that is not publicly accessible, then ask the instructor to create a private GitHub or GitLab repository for you. You will do your work in this version control repository. If you wish to use a collaborative editor such as Google Docs, for your written paper only (not for storing experimental data, for example), you may do so. Also create a mailing list for your team. Give the instructor access to all of these resources. Consider creating a catchy team or project name, to improve esprit de corps.

Exactly one member of the team should submit a single PDF, of approximately 4-5 pages, named named teamname-proposal.pdf. The proposal document should satisfy all the requirements of the project proposal — many of the submissions for that assignment did not. The proposal document should also address all the feedback you received on your proposal and presentation. If you have questions, make an appointment with the instructor.

For this submission, include a week-by-week schedule of the work you expect to complete. Ensure that each week's milestone is concrete and measurable, so that you know whether you have achieved it. (For example, “coding is 50% done” is not measurable, but “parsing module is complete and tested” is.) Don't forget to allocate time for building your experimental infrastructure. (You should think about it deeply enough to be able to give a detailed estimate.) The schedule should not leave writing to the last few weeks. Rather, you should write up your results as you obtain them. That will help you to double-check your methodology and experimental infrastructure, will cause you to notice issues that otherwise might have gone unnoticed until the end of the quarter, and will enable you to get feedback, which will streamline your work and prevent you from repeating mistakes.

As a reminder about some of the requirements, your proposal must:

• state a problem to be solved or a question to be answered (sometimes, but not always, there are several related ones). Give concrete examples to make the problem clear.
• indicate why this problem is of scientific and/or practical interest. Why is the solution or the answer not obvious? How is the problem solved today, and what are the limitations of current practice?
• indicate your planned approach, such as an outline of your algorithm or your planned experimental methodology.
• outline your experiments. What experimental subjects (humans, or more likely programs) will you experiment on? What will the experimenters do? What measurements will you take? (Be specific, not vague. Include the putative tables or graphs that you anticipate being in your final report, with the headings and/or axes labeled but no data filled in yet.) How will you know whether you have been successful?
• discuss related work, with citations to the scientific literature.