5
1

Hi,

I am a young Operations researcher and am in the field for past 3 years since the start of my graduate school at Texas A&M. I am sure many of you have been reading a lot of research papers. So I wanted to know if you guys can share with me, how you approach reading research papers? like for e.g. do you guys make notes while reading? or do you guys read it bit by bit over a few days? or you read it at a stretch in a single sitting?

Thanks, Venky

asked 12 Aug '10, 15:48

Venky's gravatar image

Venky
7681219
accept rate: 15%


  1. Caffeine. Lots and lots of caffeine!
  2. If, after the title and abstract, I think the paper is worthy of a very close reading (or if, heaven forfend, I'm reviewing it), I likely will take notes, either in Tomboy, if I can keep the notes short and nontechnical, or in LyX -- a writing program that's a front-end to LaTeX, with easy entry of formulas -- if the notes will be extensive and/or math-heavy. If I'm reading the paper just on the off chance it will stimulate some cerebral activity (or I have nothing better with me to read at lunch), I'll skip the notes.
  3. I read a lot of papers with math programming models, but this next bit may also apply to stochastic process models. Fairly often I'll find myself midway through the paper and having a hard time remembering either what a particular symbol represents or what a particular equation/formula says. I have a mix of tricks for dealing with that. I may type definitions/models into my notes, so that I can refer back quickly. If the paper is in a PDF file, I will sometimes use a PDF utility (such as PDFSAM) to extract just the pages with definitions/models in them, so that I can refer back to them quickly. (You could also just copy the document with a new name, open it side-by-side with the original, and freeze the copy on the formula pages.) For a print document, I might make a hard copy of the models/definitions for quick reference (although a paper clip/sticky note frequently works just as well).
  4. Reading an article across a span of days involves a considerable amount of setup time on each day after the first (getting back up to speed). This rework time increases over time (possibly for biological reasons, more likely due to cumulative contact with students). Doing it in one sitting is usually preferable if the paper is good enough/important enough to capture your attention. For less interesting papers, I'll read bits here and there in the "cracks" in my schedule and not bother reviewing previously read bits unless absolutely essential.
link

answered 12 Aug '10, 18:32

Paul%20Rubin's gravatar image

Paul Rubin ♦♦
14.6k412
accept rate: 19%

I cannot agree more with you on number 4.

(13 Aug '10, 06:47) Mark ♦

Thanks for sharing your strategy. I could use point 3, as that the biggest trouble I face when i am reading a paper and keep switching back and forth.

(13 Aug '10, 20:53) Venky

This is slightly related to what I have been focusing on lately. If you take away the technical side of the question i.e note taking an so forth, then you are left with the issue of how to keep 200% concentration over a longer period. I am in the process of building a new code base of several 100K lines of complicated math code. I see lots of fall outs in my concentration and distractions which hurts productivity. The advise is actually really simple, but works.

First you have to restructure your day and do the hard mind stuff early in the day and do dummy work when you are tired in the evening (night..).

I often work on several projects during a day, so I can switch project when I loose my energy on the other.

I hate to sound like a health-freak (which I am not), but workout and what you eat has an great impact on how effective you are.

I think experts would recommend a good night sleep, but it's a luxury I don't have, so I am trying to short cut here :-)

link

answered 13 Aug '10, 08:41

Bo%20Jensen's gravatar image

Bo Jensen ♦
5.0k2919
accept rate: 14%

You completely added a new dimension to the question. I couldnt agree with you more. Mornings should be reserved for a brain workout. :)

(13 Aug '10, 20:49) Venky

Newsweek recently published a cluster of articles on creativity that contained at least anecdotal evidence (possibly research, can't remember) supporting your idea of having multiple projects going at once and switching when you start to "block" on one. There's also evidence that getting up and moving around periodically (taking a walk, for instance) helps -- probably something to do with all the blood pooling in your butt if you sit for too long.

(14 Aug '10, 15:53) Paul Rubin ♦♦

I think it is a wonderful question. I really like to see what other people have to say. Obviusly my paper reading skills are not as refined as professor Rubin but here is the method that I have been using consistently (and somehow successfully) over the course of a couple of years. Throughout the day I see a lot of papers and I typically need to first determine if that is in line with my work or not. If that's something that my advisor has sent to me, I have to read it and there is no exception but if I've found the paper online I first need to see whether or not it is useful in my research.

I also try to keep three things close to me when reading, a highlighting marker, a pen or pencil (I love pencils but I only use them at my desk in University because we have an electric pencil sharpener and it makes it convenient to use :) and my notebook (I use notebooks since I tend to lose my notes if I write them in single sheets of papers. Also notebooks can help me go back in time and read my notes that I have taken 2 or 3 years ago, it is a good system I highly recommend it if your system is not yet fully digital :) I have seen many people who use latex but I don't use it.

Before starting, I also try to look at the number of citations that a paper has received sometimes I feel that a paper is simple but it has received a lot of citation (like Google's MapReduce) or it receives no citation but it has a very hyped up title (which means I should avoid the paper). Citations can help me determine if it is worth reading.

Here is how I read papers. As I said I have tried to keep it systematic and consistent over time:

  1. Title + Abstract: I read them quickly to make sure it is in line with what I want.
  2. Introduction + Conclusion: these two parts determine whether or not you need to ready the body of the paper or print it. For example I pay attention to these items: a) problem definition and problem setting b) assumptions c) what kind of model are they using d) solution approach and method
  3. The body (this is the least important part of the paper :) most OR papers have mathematical models in 1 or 2 pages I typically highlight them if I am reading a hard copy (which shows that I actually care enough that I have printed the paper) At this point I start taking notes in my notebook and mark that page with the paper title and date so I can refer to it.
  4. TODO: I wish at some point I post my notes to a blog as I am reading papers. I think many people will benefit that if we all do it. An example of my effort on Fisher's classic paper can be found here
link

answered 13 Aug '10, 06:23

Mark's gravatar image

Mark ♦
3.6k22350
accept rate: 9%

I really liked your idea of blogging a summary of the paper.

(13 Aug '10, 20:51) Venky

+1 (particularly if I can delegate the actual work to doctoral students :) Hopefully for important papers the blog comments can turn into an interchange on implementation issues, limits to applicability, etc.

Re # of citations, two caveats. The first is that, obviously, freshly published papers have no citations. The second is that citation indices do not distinguish between "Rubin's seminal paper on ..." and "the village idiot Rubin once again stepped in it with ..."

(14 Aug '10, 15:49) Paul Rubin ♦♦

One other thing -- consider using a mechanical pencil. Environmentalists might read your post and target you as a promoter of deforestation.

(14 Aug '10, 15:50) Paul Rubin ♦♦

I've recently been reading a lot of papers, researching a specific range of topics. My procedure works like this:

  • I download a PDF of the paper. Almost all my reading is done on a computer screen rather than printing each paper out.
  • I enter the citation details for the paper into my bibliographic software (I use ProCite, but would like to eventually move to Endnote). The software gives the record an id number, and I name the PDF file with that number, the journal and the year, e.g. "27650 Eur J Oper Res 2008". That lets me find it again really quickly - I just keep all the PDFs in a big folder.
  • Also into the bibliographic software I copy the abstract, and I assign it a number of keyword tags based on a quick scan of what it seems to cover. This means that when I'm searching for relevant information later, I can find things quickly, and I can do reviews of specific topics easily.
  • When reading the paper, I usually do a quick scan through first, noting mentally the structure and any interesting looking aspects. I then decide if I want to read the paper in detail. Often I won't, so I'll update any notes or a brief description in Procite and move on.
  • If I decide to read it in detail, I usually do it in one sitting. I make any notes in Procite and give a brief description of the paper as it relates to my interests, and give it a "quality rank", which is just an indication for myself of how good I think the paper is - it helps narrow down the list when doing later reviews.
  • I don't bother working through complicated maths unless the paper is absolutely relevant to stuff I'm currently working on.
link

answered 16 Aug '10, 06:00

DC%20Woods's gravatar image

DC Woods ♦
4.1k22546
accept rate: 5%

One thing to add to excellent reading techniques above, I use alert functions of journals or aggregate websites. It really helps to track relevant papers.

link

answered 20 Aug '10, 16:37

Ahmet%20Yukselturk's gravatar image

Ahmet Yuksel...
196117
accept rate: 14%

Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here

By RSS:

Answers

Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "Title")
  • image?![alt text](/path/img.jpg "Title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Tags:

×8
×1

Asked: 12 Aug '10, 15:48

Seen: 2,143 times

Last updated: 20 Aug '10, 16:37

OR-Exchange! Your site for questions, answers, and announcements about operations research.