Reading Papers
All students must read the assigned papers before class. Reading means working through a text, figuring out the key innovation in the paper, and making up and working through examples that help you understand the innovations. While reading the paper, here are some things to think about. (Bring your notes and thoughts about these to class.)- What is the contribution of the paper? How important is this contribution?
- What problem is the paper addressing? How well do they succeed in solving the problem?
- How would you have addressed the problem?
- Are the language mechanisms intuitive? Easy to use? Efficient? Scalable?
- What was unclear to you?
- How does the paper relate to other papers we have read?
- Are there any obvious or non-obvious extensions to this work?
- Can you suggest a two-sentence project idea based on the ideas of this paper?
Presenting Papers
In each class, a student will present one or two research papers. Each presenter must discuss the papers and presentation with me at least 3 days before the class.Things to think about when preparing your presentation:
- Your primary goal is to clearly communicate the key ideas of the paper. In order to do this, you may need to read some of the related work, and you may need to present background material.
- Your secondary goal is to prompt discussion about the papers.
- Aim for a talk length of 80 minutes. (Questions during the talk and discussion afterward should take up the remaining 20 minutes of class time.)
- You do not have time to cover every detail of the paper. Focus instead on the important and/or interesting aspects of the paper.
- The structure of the paper is often useful as the structure of your presentation. However, this is not always the case. Make a conscious decision about how you will structure your presentation, and why.
- If you are presenting two papers, think about how you can synthesize the work in your presentation. How are the two papers similar? How are they different? How are they related?
- The presentation should be in the format of a lecture. You may give a whiteboard lecture or a lecture with slides. (Talk to me about which would be better given the material you are presenting.)
- Do not read aloud the text on your slides. Do not use your slides as a crutch. (If you use slides, you must nonetheless be fluent enough with the material that you could do the entire presentation using a whiteboard!)
- For advice on giving good presentations, see, for example, Mark Hill's advice, Matthew Miller's tips, or Simon Peyton Jones' slides on how to give a good research talk (via Mark Leone's advice website).
Process
- Paper Preferences: By Friday, January 18th, send me a ranked list with your top 3 choices of dates/papers you'd like to present (with subject "CS7480 Preferences"). I will confirm presentation dates by Tuesday, January 22nd.
- Partners: I will assign a partner for each of you to work with. You will help each other prepare for your presentations, practice the presentations with each other, and proofread and provide comments on each other's lecture notes.
- Check In: Email me a draft of your lecture notes or slides at least 4 days before your lecture. By this point, also make an appointment to meet with me a few days before your presentation to check in with me.
- Peer Evaluation: Your peers will anonymously evaluate your presentation with the help of this evaluation form.
Critiques of Papers
Critiques should be emailed to amal@ccs.neu.edu, with the subject "PL critique 2019-MM-DD "(with month and date replacing MM and DD).Deadline: Critiques are due by 11am on the day we're scheduled to discuss the paper in class.
Your critiques/reviews should consist of a summary of the paper, a discussion of its strengths and weaknesses, and technical comparisons with papers we've read earlier in the semester (if appropriate). You can also include discussion of issues you found confusing, as well as ideas for extensions to this work and ideas for future projects. In particular, you should try to answer the questions listed in the "Reading Papers" section above. Below are some additional guidelines.
- Summary When writing the summary (1-2 paragraphs), try not to repeat sentences/phrases from the paper's abstract, introduction, or conclusion. Your summary should demonstrate that you understand the paper and are able to summarize it concisely and place the work in context. The reader who has not read the paper should be able to read your summary and find out what the paper is about. A good summary is one that is more informative than the paper's abstract!
- Strengths and Weaknesses Remember that
you are required to list at least 2-3 strengths and 2-3
weaknesses -- i.e., feel free to list more! Clearly explain
each strength/weakness (usually a paragraph or two).
Provide specifics, justify your opinions, be accurate,
and be fair. For instance, it's very easy to
criticize a paper for not yet having solved a problem in its
full generality: "the paper tackles System F, but it doesn't
say anything about how we would do this for Java!" Criticism
of that sort is somewhat irresponsible. If the paper is the first to
demonstrate how to do X for System F, assuming X is a
worthwhile problem beyond the level of a graduate student
exercise, then the paper has made a novel contribution that
should be recognized (even if the solution seems
"obvious" to you now that you've read the paper!). Similarly,
it is irresponsible to simply say that the approach "will not
scale." You have to provide a coherent justification for all
your opinions. You must explain why you believe that it
will not scale: "it will not scale because M and N showed
that foo is not possible " OR "it is
unlikely to scale because many attempts to
scale to settings with A failed in the 80s. The main
difficulty with A is how to do B, but this paper does not provide
any fresh perspective on how to tackle B."
There are several questions/criteria that you should keep in mind as you think about the strengths and weaknesses of the paper. You may want to read John Ousterhout's Hints for Reviewing Papers and pay attention to the questions listed in the "Issues" section and the "Structure" section.
Last modified: Mon Jan 7 14:21:23 EST 2019