Extract information from PDF

[alpkaanaksu]

Extract DOI (which can then be used to fetch relevant information)

Extract other relevant keywords / links (e.g. for detecting preregistration automatically)

[alpkaanaksu]

I implemented a first version (ee58a54572d078e90938870ff4e97910325712f7).

We open the PDF file, extract its text content, find all DOIs and assume that the first one is the one we are looking for.

It should be fairly easy to implement some kind of keyword/link scan. Reading PDFs and extracting text is much easier than I expected.

https://github.com/nicebread/RESQUE/assets/68744864/cb826b3d-78cc-47b1-9e1a-092d72fbd695

[nicebread]

Wow, this is great!

A minor feature request: When you imported a Pdf, the focus should already be at the new publication (and not on the "author" tab).

A general question is, how much "intelligence" we put into the webform, and how much into the subsequent R analysis script. E.g., one could attempt to fill in some fields based on automatic tools, such as:

https://github.com/quest-bih/oddpub

https://github.com/serghiou/rtransparent

(both are R-packages, but maybe there are alternatives for JS?)

[alpkaanaksu]

I doubt that there are similar packages for JS. But since they are both open source, we could implement a minimal subset of features in JS. (or maybe even the whole thing, we could publish the package)

[alpkaanaksu]

A general question is, how much "intelligence" we put into the webform, and how much into the subsequent R analysis script.

Having just enough 'intelligence' in the webform to fill in as many fields as possible automatically should be enough.

A basic principle could be: If we ask for information which can be found in the text, we should try to extract it from the PDF.

[alpkaanaksu]

When you imported a Pdf, the focus should already be at the new publication (and not on the "author" tab)

Done: d8cfa5b697ccd0cb497605027bf3aea9a1fa6e5f

[nicebread]

Maybe check https://github.com/CeON/CERMINE