[NFB-Science] Accessing Figures in Scientific Papers

[Currin, Kevin]

Hi Campbell,

For making graphs, I'm a big fan of the R statistical program. There are a lot of biology/bioinformatics packages for R as well, which makes it very useful as you progress in research. I really like python for text parsing, but I much prefer R for making graphs and actual statistical analyses. However, I know researchers who much prefer python for statistics and graphics as well, so it is all down to personal preference. There are a few ways to make R plots accessible as well that I could help with.

I don't have a ton of experience with AI tools, but I was very impressed with chat GPT's description of a figure from one of my papers. It gave a very good high level description at first and it can give more detail if you ask about specific parts of the figure. However, the free version requires the image to be in an image file, like PNG, by itself. I think there is a paid version that will parse PDF files for graphs, but I haven't tried it.

Learning what all the different types of graphs "look like" and what information they convey is very useful, and learning to produce graphics to help communicate with sighted colleagues is essential. However, there is a ton of information you can gain about a figure just from reading the results text of a paper and the figure legends. In many cases, you may not need a detailed description of the graph because the findings will be described, even if the specific parts of the graph are not. There will of course be cases where you need more info about a figure part, but it is likely you will only need to ask for sighted assistance with specific parts, which will save a lot of time over getting the entire figure described or in an alternative format. I've noticed that some of my colleagues only look at figures and do not read the results text of papers very closely. Therefore, I often pick up on findings that they might miss because they aren't directly in the main figure (it could be in a supplemental figure or table, or simply just described in the text). There are many papers I have read where I didn't need any assistance with figures because the text was clear enough. This will definitely become easier as you progress in research and have an idea in your head about how different types of data is displayed in your field.

When you do need to have figures described, I've found the draftsman tactile drawing kit very useful. It allows both you and a colleague to draw raised line diagrams on special film paper that you can save for later reference.

Thanks,

Kevin
________________________________________
From: NFB-Science [nfb-science-bounces at nfbnet.org] on behalf of Robert Jaquiss via NFB-Science [nfb-science at nfbnet.org]
Sent: Sunday, June 2, 2024 10:02 PM
To: 'NFB Science and Engineering Division List'
Cc: rjaquiss at earthlink.net
Subject: Re: [NFB-Science] Accessing Figures in Scientific Papers

Hello Campbell:

     If the figures are black and white line art, it should be fairly easy
to print them on Swelltouch capsule paper. A sighted person will likely need
to enlarge the figures. SwellTouch paper is available from American
Thermoform Corp. SwellTouch paper comes in three sizes; 8.5X11, 11X11.5 and
11X17. It can be printed on using a copier or laser printer. The paper is
then processed with a machine such as a SwellForm machine.
Note, only black lines will raise on either SwellTouch paper or PIAF paper.
If you have more questions, call me at 909-593-6711, Ext.102.

Regards,
Robert



-----Original Message-----
From: NFB-Science <nfb-science-bounces at nfbnet.org> On Behalf Of Campbell
Rutherford via NFB-Science
Sent: Saturday, June 1, 2024 1:07 PM
To: nfb-science at nfbnet.org
Cc: campbell.rutherford15 at gmail.com
Subject: [NFB-Science] Accessing Figures in Scientific Papers

Hello all,

I am about to begin my first summer research experience, which will mostly
involve computation and data analysis. One unanticipated problem that I
encountered was that of figures without alt text in scientific papers. Note
that these figures appear in PDFs that are otherwise accessible. I've been a
bit spoiled so far in college; if figures in the papers that I had to read
were not already accessible, one of my accommodations was for someone to
write descriptions for me that would include the necessary data. However, in
this research position, the resources to make this happen are not already in
place, and the scientists with whom I will be working and I are trying to
figure out how to best make figures in scientific papers accessible to me.

Has anyone had any luck with using gen AI or other technology to describe
figures? If so, what is your app/program of choice? Also, what is the best
way to produce figures from your own data? I have used Desmos for graphs of
functions and Excel for simple statistical graphs (line charts,
scatterplots, bar graphs, etc.). I also played with SAS Graphics Accelerator
a few years ago, and one of the scientists with whom I'll be working says
that he typically creates his figures using Python. What do you all think
works the best? Are there other methods for creating figures of which I
should be aware?

Thank you,

Campbell Rutherford

_______________________________________________
NFB-Science mailing list
NFB-Science at nfbnet.org
http://nfbnet.org/mailman/listinfo/nfb-science_nfbnet.org
To unsubscribe, change your list options or get your account info for
NFB-Science:
http://nfbnet.org/mailman/options/nfb-science_nfbnet.org/rjaquiss%40earthlin
k.net


_______________________________________________
NFB-Science mailing list
NFB-Science at nfbnet.org
http://nfbnet.org/mailman/listinfo/nfb-science_nfbnet.org
To unsubscribe, change your list options or get your account info for NFB-Science:
http://nfbnet.org/mailman/options/nfb-science_nfbnet.org/kwcurrin%40email.unc.edu