General Info

CPSC 502: Research Project - Assignments

In this course, we will have five assignments:

Before going into these assignments in detail, here are some general remarks. The first four assignments are papers that a student has to hand to his/her supervisor and the instructor. "Hand to" means that the student emails a pdf-file to both, supervisor and instructor. Note that only pdf-files will be accepted, no Postscript-files and definitely no Word-documents!

Both, papers for conferences (or journals) and research proposals to funding agencies have to adhere to rules regarding the formatting of the paper and for this course we also require that every document you produce for the assignments adheres to a particular style, namely the ACM SIG style for articles in proceedings.

Plagiarism has become quite a problem in academics, as is indicated by the fact that organisations like IEEE have now guidelines how to deal with plagiarism. While the instructor is not aware that there were any problems regarding plagiarism in CPSC 502 in the past, it is nevertheless recommended that the students take a look at Saul Greenberg's page on plagiarism.

Some criteria for grading that will be applied to all papers submitted are

  • Understandability of the English
  • Correctness of the grammar
  • Quality of figures and tables (if there are any)
  • Correctness of the formatting and compliance to the rules for the assignment (the first is mainly for those insisting on letting Microsoft decide how their documents look like, which can be a bad mistake)

While for all papers these will not be the main criteria, it should be clear that papers that are difficult to understand (due to how they were written) and terrible to look at will not get good grades!

Research proposal:

The purpose of the research proposal is to give its reader an idea what research the project described by the proposal is supposed to perform. This means that there need to be at least two things in such a proposal:

  • A description of the problem that the project wants to address/solve, done in a way that even a non-expert in the field (but you can assume someone with a general Computer Science background) can understand.
  • An initial idea how this solution might look like. This might not be what in the end will be the solution presented in the final report and presentation, but it should become clear that there is at least a starting point towards solving the problem.

Additionally, most proposals will already list some related work in order to provide a basic understanding what the state-of-the-art in solving the problem is and where the proposal differs from this state-of-the-art. Citations of related work should also be there to point to in-depth descriptions of the problem (if such descriptions exist).

The proposal should not be more than 3 pages!

Deadline: October 3, 2008; noon

Related work list with 2 critiques:

Very seldomly research works in a vacuum. Instead, there are usually many research results that are related to the research one is doing, although, naturally, none of these results should be what the project tries to accomplish. Therefore it is very important to become aware of all the research related to one's own and this assignment will require the students to do exactly that.

While the required list is exactly what it says: a list, the term "critique" requires some explanation. The following was written by Denilson Barbosa for a previous version of this class and is copied here with his permission:

The purpose of a critique is to get you to start doing some research and also to see how other people have written up their research. Hopefully you will notice that some research papers are quite difficult to follow while others are made clear and readable. Most are a mixture of the two. Remember this is an objective task, sentences saying : "I feel that ... " have no place in a scientific paper, your feelings are irrelevant, only objective criticism, which you can support should be recorded in your work.

As a suggestion, organize each critique into three sections as follows:

  1. Overview: 1 paragraph stating clearly and objectively what is in the paper (an empyrical study? a new theory? an algorithm?). Most papers will have both a theoretical framework and an experimental evaluation.
  2. Strengths: 2 or 3 paragraphs stating clearly and objectively the strong points in the paper. Start by focusing on the following questions:
    • is the work relevant? (does it address a real problem?)
    • is the theory sound?
    • are the experiments carefully done?
    • are the results presented in a meaningful and clear way?
    • what is the impact of the paper? will it change the way we do things?
    • is the paper well written?
  3. Weaknesses: 2 or 3 paragraphs stating clearly what could be improved in the paper. Revisit the questions above and find those points that are lacking in strength. Keep in mind that while there is no perfect paper, if you dislike everything in a paper you are probably either being too harsh or reading irrelevant work.

Deadline: October 17, 2008; noon

Interim report:

A little bit over halfway through this course, a student should have a rather good idea how he or she wants to solve the problem that he/she set out to solve in the project proposal. Therefore, in the interim report, the student should be able to present this solution idea together with a plan how to evaluate that the idea really works (for some areas, the later is no big problem, for other areas quite some research has to go into this plan).

A general structure for an interim report is as follows:

  1. Abstract
  2. Introduction/Motivation
  3. Basic definitions/Explanation of the problem
  4. Proposed solution to problem
  5. Related work
  6. Intended evaluation of proposed solution
  7. Conclusion
  8. Bibliography
The section titles have to be seen more as an indication what should be in the section and they can be changed to fit better to the particular project. Also, where the related work section goes can differ from project to project (and sometimes this section even has to be merged with the explanation of the problem section).

The interim report should not be more than 7 pages!

Deadline: January 16, 2009; noon

Final report:

A key point in research is to document the results of the research, so that others can learn from these results and build upon them. Usually, reports on research are done in the form of papers that are submitted to conferences or journals (and are hopefully accepted there). In most areas of Computer Science the main venue to send such a paper are conferences (and some of the research done for this class might end up at a conference) and it is usually very difficult to get a paper into one of the top conferences of an area, because a lot of people submit papers and there are only a limited number of papers that can be presented at a conference. Therefore the review process is very hard and just having one weak accept (and all other reviews suggest a strong accept) might not be enough to get a paper in (and having one or several reject will result in a rejection of the paper).

In this class, we will "accept" all final papers submitted by the students, but their quality will be reflected by the grade you get and you will naturally get a "review" (justifying the grade) from the instructor.

The structure of the final report should be similar to the interim report, except that now there is no section on the intended evaluation but a section that, in fact, presents the evaluation of your solution. For some students it might be possible to reuse quite a bit of the interim report (but be careful with some of the phrases: in the interim report you might write something like "we intent to ..." which naturally you must have done at the end of the class), while for others the additional research and evaluation might have led to some re-thinking of things and this naturally has to be reflected in the final report.

Not every idea for solving a problem really works. Therefore it can be the case that the result of a student's research is that a particular idea does not work (demonstrated by the evaluation). While this is usually difficult to submit to a conference or journal, it is acceptable for this class, if the report makes clear that the failure is due to the initial idea (and this could only be seen by going all the way to the evaluation) and not due to the student (i.e. there is literature that indicates that the basic idea is not working well and the student did not find it or the evaluation is very sloppy, perhaps one example, so that it is not even clear that the method does not work, and so on). A well-written final report indicating why the initial idea does not work as expected (supported by experiments or sound reasoning) can get as good a grade as a paper reporting on a success.

The final report should not be more than 10 pages!

Deadline: April 3, 2009; noon

Final presentation:

The end of a successful research project is usually the presentation of the results in front of the audience of a conference in which a paper about the research was accepted. The final presentation will try to mimic that. All presentations will be announced to the Department and everyone is welcome to sit in these presentations (even people from outside of the Department). We will schedule each presentation so that at least the supervisor and the instructor can attend it.

The presentation should

  • give an introduction to the problem that the research tried to solve,
  • report briefly on what the state-of-the-art was before the research took place,
  • present the solution idea developed by the student, and
  • present an evaluation of the results produced by the student.
So, the general structure is very similar to the structure of the final report.

Most students will use slides in their presentation and when we schedule the presentations each student should make sure to tell the instructor what equipment he or she needs to have in the presentation!

There are several different ways how to do good presentations and usually it takes a researcher some time to figure out what way is best for him or her (for some researchers this time even seems to converge against infinity). The materials page contains several links with tips for presentations.

The length of each presentation should be not more than 30 minutes and there will be a question period of also up to 30 minutes after the presentation.

Will be scheduled: April 13 to 17, 2009

to the general timetable for the course.

Last Change: 1/8/2008