How to read a lot of research papers?

Question

TL; DR: At least for the next few weeks, I'm going to be overwhelmed with reading a lot of research papers one after the other, sometimes even at the same time. How to deal with this? Are there any tricks to reading many research papers simultaneously?

---

I am a senior year undergraduate in Computer Science.

This semester, I took three of research-based courses in Computer Systems.

1. A course named "Topics in YYY"
2. A seminar course
3. A UG Research Project course [This is essentially "my" project, with a professor and a PhD student to advise and support me]

In 1, I am usually expected to read 2 papers per week.
In 2 as well, I have to read, review and present 2 papers per week.
For 3, initially there was less reading and more of brainstorming and experimentations. However, now that we're going to be writing a research paper based on our initial findings, there are a lot of "related" work papers that I have to go through.

In 1 also, initial reading was easy, but as the course is progressing the readings are getting advanced.

This is probably a result of poor planning on my part. I should have taken some programming-related or basic courses instead.

But the end result is, at least for the next few weeks, I'm going to be overwhelmed with reading a lot of research papers one after the other, sometimes even at the same time.

I could drop either 1 or 2 but that means running away from the responsibility; and my transcript will show the withdraw grade, so that's not really cool either.

How to deal with this? Are there any tricks to reading many research papers simultaneously?

Answer 1

There probably is no way of developing a speed-reading skill on such a short notice. You could approach the papers in a structured way.

Have a table (with more rows than shown) of data that you fill out for each paper:

Aspect	Paper 1	Paper 2
Method
Novel results

You can consume the papers as a categorisation task where you look for the relevant information which you can report and forget or leave for later reading. The task of reading a whole paper is replaced with filling a table with data and moving on. Obviously this is not the same as reading and understanding a paper but would make your current situation manageable.

You don't need to fully comprehend every paper but will still become familiar with them and understand where further reading is needed.

A good order of reading a paper would then be: Introduction -> Conclusion -> Results -> Methodology. Literature review can be skipped unless you are interested in context.

Answer 2

My suggestion might turn out to be inappropriate to your domain of expertise but in my own area(s) it has proved very useful. Moreover, my method has nothing to do with so-called speed reading.

Many papers are published as essentially standalone entities. The authors assume a certain vocabulary on the part of the reader in a particular field, but do not assume that the reader knows very much about the sub-field in question. As a consequence, the introductory parts of the article, which might be several pages long, provide history, background and motivation for the research. Finally, after presenting the method and results, another long section is devoted to the discussion.

Basically, I skip the introduction and discussion in a great many papers and proceed instead, directly to the method and results. I know enough of my research areas to feel reasonably confident about my own assessment of the usefulness and validity of various research methods (notwithstanding that they have already survived the scrutiny of peer-review). If the methods survive my own assessment, I go ahead and read the results. Then, rather than reading what is often a long discussion that recalls many aspects of the introduction, I spend time thinking. To my mind, thinking often pays dividends that far surpass the time spent reading. To be more specific, I spend time considering whether the results pass a simply plausibility test and what kind of conclusions I can draw from the results in light of the various limitations (and of course positive aspects) of the methods.

And that's it! I'm not, of course, suggesting that you never read the introductory parts of a research paper, or the discussion. Rather, I'm suggesting that you can use your time more effectively by reading the parts of a paper that are most likely to tell you something new.

Having written this, I realize that my suggestion is perhaps least applicable to mathematics. Mathematicians seem to have a knack of getting to the meat of a research paper without too much flim-flam at the front. The results, instead, are allowed to speak for themselves.

Answer 3

I share with you something that I already posted in this conversation:
How to write a "Related Work" section in Computer Science?

Perhaps how to read scientific paper faster (taking into account IMRaD structure) is the article that will interest you the most. However, this is what I usually suggest to my students. You may want to start from this link to my website contains a selection of videos that can help you in conducting a related work, depending on where you are in your journey.

You also may want to consider the following tips for reading, selecting and categorizing the works that you want to include:

• Not all papers are made equal: heuristics for understanding if a paper is worth reading or you should move forward to the next one.
• Not all journal and conferences are made equal: understand how to select good venues (conference or journal) where you can search for good publication.
• Quantity may be important: learn how to read a scientific paper faster and more effectively.
• Keep your resources in an orderly manner: master the right features in MS-Word for handling the related work and managing the growing complexity of the task.
• A literature review is about put some order in the chaos: understand how to categorize the papers that you are reading. In this way it will be beneficial for your readers.