How do you scale up your scientific reading?

[Ivan Stepnov]

Everyone knows that the number of publications is growing every year, this is especially noticeable in biology. I'm wondering how other biologists cope with this volume. Is it possible to scale up reading papers? It is clear that some papers need more thoughtfull process like review papers and breakthrough papers. I think that especially scientists from the industry should know about such tricks to scale things (reading a lot in a tight timeline). What do you use in your pratice?

[CharonY]

I found to rely on keywords more than, say two decades ago (where I would read whole journals more often). When writing reviews I often use pubmed to create a document with all the abstracts and tag those that warrant a closer read.

Among those I check intro and maybe a bit of the discussion to see if it is relevant to my topic. In contrast to some of my colleagues I try to avoid relying on known authors too much to avoid biasing against newer or less well known authors (but they can be a positive control yo make sure my search strategy did not suck). But overall it is true that it is difficult to keep track of things.

[StringJunky]

16 minutes ago, CharonY said:

I found to rely on keywords more than, say two decades ago (where I would read whole journals more often). When writing reviews I often use pubmed to create a document with all the abstracts and tag those that warrant a closer read.

Among those I check intro and maybe a bit of the discussion to see if it is relevant to my topic. In contrast to some of my colleagues I try to avoid relying on known authors too much to avoid biasing against newer or less well known authors (but they can be a positive control yo make sure my search strategy did not suck). But overall it is true that it is difficult to keep track of things.

Tha's how I would it if I was doing that sort of thing and try to find keywords that are as unique to my query as possible to aid filtering. If a research area has a specific 'lingo' it might help to use those as keywords as well.

Edited March 28, 2020 by StringJunky

[Ivan Stepnov]

57 minutes ago, CharonY said:

I found to rely on keywords more than, say two decades ago (where I would read whole journals more often). When writing reviews I often use pubmed to create a document with all the abstracts and tag those that warrant a closer read.

Among those I check intro and maybe a bit of the discussion to see if it is relevant to my topic. In contrast to some of my colleagues I try to avoid relying on known authors too much to avoid biasing against newer or less well known authors (but they can be a positive control yo make sure my search strategy did not suck). But overall it is true that it is difficult to keep track of things.

It's a viable approach. But why do you read discussion after intro (actually most of the time just last paragraph from intro)? I thought materials and methods is the best way to understand if the paper's study is aligned with my research.

[CharonY]

Depends on what I am writing. For a technique paper I'd go for methods. Otherwise intro and discussion provide context and relevance.

[Ivan Stepnov]

Keywords are great, but they lack context. Say you want find if A is related to B.  You will have a search with the keywords which will hit all documents having any reference to A and B. Next you have to pass through all links and read if there is any real relation. No context.
 

Edited March 28, 2020 by Ivan Stepnov

[CharonY]

The next step is usually generating a list based on the references in the initial screen and expand the search. But abstracts give me a good sense whether I got good coverage. Obviously one also need to apply ones own expertise. For students or laypersons a guided approach is usually better.

[Ivan Stepnov]

5 minutes ago, CharonY said:

The next step is usually generating a list based on the references in the initial screen and expand the search. But abstracts give me a good sense whether I got good coverage. Obviously one also need to apply ones own expertise. For students or laypersons a guided approach is usually better.

interesting. What do you mean by expand search? My expectation is to narrow it down at this point.

[swansont]

The number of journals has been going up, but then so have the number if subfields because of new discoveries. There really aren't more journals in my area of specialty.

[CharonY]

12 hours ago, Ivan Stepnov said:

interesting. What do you mean by expand search? My expectation is to narrow it down at this point.

I do several sweeps, usually. The first is as described above. Then I often (unless it is a narrow focus topic e.g. in a mini-review) want to expand the scope and provide a bit of synthesis from other viewpoints. Here I often use the discussions I read or sometimes ideas in my head to perform another round of expanded search, but often with a special focus. E.g if the review is about a certain host-pathogen interaction and their peculiarities, I might check other pathogens (or hosts) and contrast these interactions and discuss similarities and unique aspects.

But you are right, if there are time constraints I might focus in just getting the core papers. I do find it less satisfying, though.

[CharonY]

On 3/28/2020 at 9:13 AM, swansont said:

The number of journals has been going up, but then so have the number if subfields because of new discoveries. There really aren't more journals in my area of specialty.

I think it depends a lot on the discipline and also how specialized one is. My work is, unfortunately, somewhat multidisciplinary, so my approach is likely very different from someone who works in a far more specialized area with more focused papers/reviews.

[swansont]

9 hours ago, CharonY said:

I think it depends a lot on the discipline and also how specialized one is. My work is, unfortunately, somewhat multidisciplinary, so my approach is likely very different from someone who works in a far more specialized area with more focused papers/reviews.

Absolutely. I do applied R&D is a very specialized area. For someone closer to the cutting edge of basic research, the landscape likely changes much more rapidly.

[Prometheus]

I wonder if machine learning could be leveraged to help academics with this problem - it's scary how well youtube algorithms know me. I think ResearchGate but it doesn't seem too sophisticated.

[iNow]

2 hours ago, Prometheus said:

I wonder if machine learning could be leveraged to help academics with this problem

It obviously depends on which institution and which problems we're addressing, but it already is being used in these ways by a growing number of analysts and researchers. 

Here's one paper among what I"m sure are many on the subject: https://royalsocietypublishing.org/doi/10.1098/rsta.2019.0054

[hypervalent_iodine]

13 hours ago, CharonY said:

My work is, unfortunately, somewhat multidisciplinary, so my approach is likely very different from someone who works in a far more specialized area with more focused papers/reviews.

I think my approach with reviews and reading papers largely mirrors yours for the same reasons, although I tend to use SciFinder and Google Scholar rather than PubMed. Usually after the first pass through to filter out obviously irrelevant papers, I download all of the references + abstracts and then read enough of the papers to be able to categorise in EndNote before I move on to reading them more in depth as I write. I can also look up papers related to particular chemical structures or reactions using structure searches, which is very helpful in capturing as much of the literature as possible, although it doesn't always work depending on the topic. 

More broadly, I find that subscribing to certain journal email lists in your field is a good way of keeping an eye on what's happening in the field. I don't always scan every email I get, but I will usually look through the list of online preprints from journals like J. Med. Chem. or J. Org. Chem. and read the papers I think look interesting. Admittedly, unless I really want to know something specific about a protocol, I often skip the methods section. 

[Prometheus]

3 hours ago, iNow said:

Here's one paper among what I"m sure are many on the subject: https://royalsocietypublishing.org/doi/10.1098/rsta.2019.0054

I meant specifically for recommending research papers to academics in a similar way that youtube videos, or whatever, are recommended (your link doesn't seem to address this, at least in the abstract). I was thinking of something more like this.

[iNow]

4 hours ago, Prometheus said:

I meant specifically for recommending research papers to academics in a similar way that youtube videos, or whatever, are recommended (your link doesn't seem to address this, at least in the abstract). I was thinking of something more like this.

Thank you for clarifying.

[hypervalent_iodine]

6 hours ago, Prometheus said:

I meant specifically for recommending research papers to academics in a similar way that youtube videos, or whatever, are recommended (your link doesn't seem to address this, at least in the abstract). I was thinking of something more like this.

A lot of journal websites do this, as does (I think) PubMed. I’m not sure how good it is, but it exists.

[CharonY]

Yeah, researchgate, Google scholars and probably a couple others do have that. For the most part I have more success with focused searches. But I also change my searches as my thinking on the problem evolves.