CS 574: Paper Summaries

An important component of this class will be the careful reading and analysis of the papers to be presented. You are required to write a summary of each paper, which will be due at midnight before the day the paper is to be presented. The summaries are to be written on a web-based form; there will be a link from the course calendar to the summary form for each paper.

Unfortunately, due to the number of people in the class, it isn't practical for me to grade all of the summaries. I will grade the first three, so you can get an idea of what I'm looking for early. I will grade additional summaries, to be picked throughout the semester (as you will see when the papers are assigned, there will be a number of papers on each general topic to be covered. I'll be grading one summary from each topic, but not telling you in advance which it will be). In the case of reviews that cannot be graded, I will keep track of whether they were turned in; this may have an effect at the end of the semester if you are very close to the border between grades.

In filling out the summaries, here's what I'm looking for on the form:

(1 point) What is the topic of this paper?

What is the paper about? For instance, "this paper discusses the Linux Second Extended File System." I have actually added this question this semester as a result of the number of people who answered the next question with something more appropriate to this question.

(2 points) What is the problem the author is trying to solve?

What, specifically, is the problem being addressed by the author? An example would be, "existing Linux file systems were severely limited by maximum file name length, file size, and file system size. They also didn't support full Unix file system semantics (particularly regarding time stamps) and had low performance" Note: a frequent mistake is answering this question with something of the form, "the author is writing a paper about...." I want to know about the work here, not the paper.

(2 points) Why is this problem important?

Who cares? Why did anybody feel like they should address the problem? This can vary from commercial reasons (a frequent reason for new processors in CS 573), to purely theoretical issues.

(3 points) What is the author's basic approach?

Here is the core of the summary: how did the author approach the problem? What are the essential features of the solution?

And again, I don't want an outline of the paper here. I want you tell me about the research.

(1 point) Is this approach successful?

Time to evaluate: does the work meet the author's stated goals?

(1 point) Where should this research go next?

Time to be original: does this paper end up suggesting anything to you regarding future work? I prefer answers that don't come out the the "Future Work" section of the paper.

The remaining questions are more for my curiosity, as I refine the selection of papers every semester.

More Notes

As you know, there has been quite a bit of discussion related to plagiarism in the past year, with an emphasis on enforceable definitions and demanding proper attributions. When summarizing a paper, it is expected that the ideas in your summary will have come from the paper you're summarizing. So, there is no need to worry about citations or other attributions; it will simply be assumed.

At the same time, please be aware that one of your primary goals in writing these summaries is to convince me that you have read and understood the paper. Cut-and-pasting sentences from the paper, or answering questions in the summary with lists of terms that appear in the paper, don't do much to convince me you know what you're talking about. I'm going to be much stricter this semester than in the past on taking points off for summaries like this.

Here's an example. One of the papers that is frequently assigned in CS 573 is a paper on network routing by Glass and Ni, whose abstract begins, "we present a model for designing wormhole routing algorithms that are deadlock free, livelock free, minimal or nonminimal, and maximally adaptive." I've lost count of the number of summaries I've received who describe the topic of the paper as, "we present a model for designing wormhole routing algorithms that are deadlock free, livelock free, minimal or nonminimal, and maximally adaptive." Some of the summaries even say "we", not "the authors." When I see a summary like this, particularly one in which the entire summary consists of similar snippets of the paper like this as the answers to all the questions, I end up with very little reason to think the student did anything more than look for likely keywords and copy them into the summary form.

Much, much better would have been a summary that described the topic as, "the paper examines the requirements for wormhole routing algorithms, deriving a class of algorithms based on disallowing certain 'turns' in a packet's path from source to destination." This topic description sounds like it was written by somebody who at least has read the paper!

Last modified: Wed Sep 13 08:20:45 MDT 2006