Paper-review assignments for this class
Contents
William's Idea
One idea is for students write up K reviews of a recent paper (say last two years) related to ASM. The format review would be like a conference paper review - summary, strengths and weaknesses and possible impact - and a recommendation for whether or not to assign the paper as required/optional reading in later classes. The reviews would not be published to the class.
For each paper reviewed, we'd set up a class-accessible wiki page where users would record a "study plan" - what background materiel they needed to read, or would recommend for others. I'm thinking we'd keep this simple, as an indented list with links and annotation, but use structured links to link it to a page for the paper reviewed, so you could find it. For instance a "study plan" might be:
Study plan for user:wcohen's review of Pang et al, ACL 2003:
- I read Turney, ACL 2002 for background, since the paper discusses this for motivation.
- I read the wikipedia page for SVMs for background.
- I read about what a linear kernel is to help understand the SVM page.
The constraint would be: only include things that were useful as 'prerequisites', use an indented list to indicate the dependencies, and have only one link per line. You'd be encouraged to link to pages in the class wiki or wikipedia if you can, otherwise to an ACM DL or ACL DL link, otherwise to anything. We could extract some maybe-useful data from this - or at least, get an idea if students find it useful to have such data.
The goal would be to make it easier for later students to do reviews. To encourage this, you would get some sort of partial credit for a writeup that made use of someone else's study plan, and would get some partial credit if someone else used one of your study plans.
The idea is partially motivated by the research problem of learning prerequisites
Give feedback on this idea
Give us your feedback - will this work? Just add some text here. You can sign it with four ~'s in a row, which will look this: Wcohen 21:13, 10 September 2012 (UTC)
- Maybe before actually hand in the assignment, students can get one round feedback from TAs or Professor so as to see they are not totally off the track. It may be useful for students who are not too familiar with this domain like myself. --Dzheng
- Publishing the summary part of a student's review along with the study plan on the wiki might be helpful for others. Zsheikh 00:01, 17 September 2012 (UTC)
- This should work, but it would also help to emphasize what practical problems each paper addresses. Because this field has a large focus on applications, it would help the class to know what problems are being addressed in the current literature, to give us ideas for our own projects. Gmontane 00:43, 17 September 2012 (UTC)
- Similar to the point above, there should be a categorization of all the papers reviewed so that we (and later students) can easily navigate and grasp the overall field. Norii 02:01, 17 September 2012 (UTC)
- I second the idea of publishing the summary part of the review. That would certainly help. Nkatariy 02:31, 17 September 2012 (UTC)
- I think it is a good idea. And for each paper, it might be useful to define unified expressions to describe the relations between paper and its main references (For example, Application, Baseline, Extension, Different point-of-view, etc.) Bliu1 02:56, 17 September 2012 (UTC)
- I think this will work provided students make an informed choice. If different people choose to review different topics it will be a good resource for the current students as well as future students to get an overall picture of the subject. Rgkulkar 03:10, 17 September 2012 (UTC)
- I think this will work and we can also write how those papers are related to our projects or the lectures so that future students can have a better view of the class. --Zeyu Zheng 03:34, 17 September 2012 (UTC)
- I think that reviewing recent papers is a good idea, especially if these papers are relevant to the class project. Ytsvetko 17:35, 17 September 2012 (UTC)
- I would like to have an overview from Professor about main research directions/ subtopics in ASM then each student/ group of students write up a short survey about one Tuan Anh 21:26, 17 September 2012 (UTC)
- Sounds like a good idea for me. My only concern is that for a student who is not really familiar with SMA, the scope of SMA (and the choice of papers) would be a little confused. Kenneth Huang 23:38, 17 September 2012 (UTC)
- It would be helpful if each student can choose papers to review. Because each student eventually would do some his/her own work in the class, it would be better if papers for review align with the student's interest so that the student can exploit the experience for the project later. Nozomi Nori 01:13, 18 September 2012 (UTC)
- Besides writing the summary, if one has read 2 papers, which addresses the same problem, one can also very briefly state the pros and cons of the methods used in the papers. --Anikag 04:05, 19 September 2012 (UTC)
- The "prerequisites" idea is great. I think that allows for the development of a great resource for later readers. I also second the idea of reviews, but would like to have these reviews published to the class. I think what we should aim to achieve from this activity is to perform an organized(and extensive) selection of papers and generate summaries not only for future readers but also current students so we all can learn from each other effort. To ensure summaries are consistent we can have 2 students collaborate on a review. --Srawat 04:18, 19 September 2012 (UTC)
Modifications to discuss on Thursday
- Clear papers not explicitly suggested on the course page with William, Aasish, or Bhavana
- Add structured links (like problems addressed, methods used, ...) as we did in previous classes
- Publish summary ? as we did in previous classes
- Ask/allow students to do overview projects/survey papers - maybe instead of a project?
Add another idea
What sort of assigments would you design? Remember, the constraints are
- Students collectively create something useful
- Completing the project moves you toward understanding an area that you might do a project on.
If you'd like to suggest something add it to this page, below.