Streamlining creation of a list of co-authors and their institutions from a bibliography of your publications
For NSF proposals it is required to provide a list of collaborators and other affiliations (COA) in an Excel file, https://www.nsf.gov/bfa/dias/policy/coa/coa_template.xlsx. And getting a list of all your recent collaborators can be a serious pain in the neck. Here is some stuff I've written that makes it a lot easier, even if it's a bit kludgy.
Author: Philip Resnik, University of Maryland (resnik@umd.edu)
If you don't use BibTex, our friend ChatGPT will make it easy for you to take whatever format you have your citations in (e.g. copy/paste from your c.v.) and create the .bib file. Just use chat.openai.com (ChatGPT 3.5 is fine) with the following prompt:
Create bibtex entries for the following papers:
<copy/paste your references here>
Depending how many refs you've got you may need to do this in multiple batches since ChatGPT has limits on the size of any single prompt.
Make a copy of authorindex.tex and follow the instructions in
the header. If you need to sort your .bib file chronologically so you
can easily copy/paste just the entries from the last five years into a
fresh .bib file to use as input, go here:
https://flamingtempura.github.io/bibtex-tidy/index.html?opt=%7B%22modify%22%3Atrue%2C%22curly%22%3Atrue%2C%22numeric%22%3Atrue%2C%22months%22%3Afalse%2C%22space%22%3A2%2C%22tab%22%3Atrue%2C%22align%22%3A13%2C%22blankLines%22%3Atrue%2C%22sort%22%3A%5B%22year%22%2C%22month%22%2C%22author%22%2C%22key%22%5D%2C%22duplicates%22%3A%5B%22key%22%2C%22doi%22%2C%22citation%22%5D%2C%22merge%22%3A%22combine%22%2C%22stripEnclosingBraces%22%3Afalse%2C%22dropAllCaps%22%3Afalse%2C%22escape%22%3Afalse%2C%22sortFields%22%3A%5B%22year%22%2C%22month%22%2C%22day%22%2C%22author%22%2C%22title%22%2C%22shorttitle%22%2C%22journal%22%2C%22booktitle%22%2C%22location%22%2C%22on%22%2C%22publisher%22%2C%22address%22%2C%22series%22%2C%22volume%22%2C%22number%22%2C%22pages%22%2C%22doi%22%2C%22isbn%22%2C%22issn%22%2C%22url%22%2C%22urldate%22%2C%22copyright%22%2C%22category%22%2C%22note%22%2C%22metadata%22%5D%2C%22stripComments%22%3Afalse%2C%22trailingCommas%22%3Afalse%2C%22encodeUrls%22%3Afalse%2C%22tidyComments%22%3Atrue%2C%22removeEmptyFields%22%3Afalse%2C%22removeDuplicateFields%22%3Afalse%2C%22lowercase%22%3Atrue%2C%22backup%22%3Atrue%7D
a handy online tool that lets you paste in a .bib file and do nice things like de-dupe, normalize entries, and sort by year. (The parameters in the url default to settings that should be good for all that, but you can customize on the page.)
Once you've followed the directions in authorindex.tex, you'll now
have a file with one co-author per line in authors.txt in lastname,
firstname format. Don't forget to delete the line containing
yourself.
Institutions are not required in the NSF COA template, but it's a nice thing to have..
python fetch_scholar_info.py < authors.txt > Collaborator_institutions.txt
Yes, that output file is weirdly in 'firstname, lastname' format. I made a mistake when specifying what I wanted to ChatGPT and decided to just live with it since this is only an intermediate file anyway.
Note that fetch_scholar_info.py uses system calls to curl. That's
because the original version using the request package led to Google
blocking the IP address; oddly with curl this was not a problem.
Note that the script takes an optional --sleep N options to sleep N
seconds between hitting Google Scholar to avoid rate limiting or
blocking. Defaults to 10 but 5 seems to work just as well.
The Google Scholar profile pages can list multiple matches, especially for common names. You'll want to manually review/edit so each author has just one line.
Sometimes the info pulled back from Google Scholar includes both a position and an institution, e.g. "Professor, Univ of Maryland". To pull out just the institution and join that info with people's names as they appear on your co-author list:
python join_scholar_info.py authors.txt Collaborator_institutions.txt > coa_info.csv
The code isn't perfect, e.g. for institutions you might see the department and institution together. It's good enough that it's not worth it (to me) to mess with trying to get the code perfect. I used CSV rather than XLSX so it would be easy to fix things in a text editor if that's preferred over editing in Excel.
The fetch_scholar_info.py script created a temporary directory for the HTML author profiles pulled from Google Scholar. Since that can take some time, to be cautious the code doesn't delete that temporary directory. At this point you should be able to go ahead and delete it.
At this point either you've got authors.txt (if you didn't add
institutions), or coa_info.csv (if you did). Either way, you can
open the file with Excel and then just copy/paste info into the NSF
COA Excel template.
Voila!