ReferenceCheck is an important tool for guiding and educating newcomers on the necessity of adding reference to the statements they make. For ReferenceCheck to be even more effective, we need to improve the tooling to which it funnels contributors into.

As Citoid's automatic reference maker is the first touchpoint for users after being prompted by ReferenceCheck we need to improve its capabilities and make it more robust so that users don't have to resort to creating a manual (basic or templated) reference. If this happens users will be faced with an unintitive user experience that leads to hightened cognitive load. Also increases the likelihood that contributors will either add no reference to their edit or abandon it altogether.

This investigation is to understand the feasibility of enabling Citoid to support PDF links and turn the bibliographic information associated into proper references.
The hypothesis is that this one of the missing capabilities needed to increase the likelihood of contributors – particularly newcomers– successfully adding a reference.

Note: Some pdf links do work. This is the case when the site returns html. If the PDF is served in a viewer or provided as a downloadable file, this is when we cannot create a formatted citation, return an error and prompt the user to create a citation manually.

Some background information:

• T136722: Get metadata from PDF links in citoid
• https://github.com/zotero/translation-server/pull/59
• Zotero translators https://github.com/zotero/translators

What we want to ivestigate in order to determine the next steps:

Data points:

• Get data on pdf URLs that are input into Citoid and how often this fails.
• Get data on number of references that include a pdf url from existing articles. We assume these will mostly be article/papers.
• Get list of PDF sites that are being inserted into Citoid by contributors
• Superset charts
• Citoid logstash

Status quo:

• Understand problem: Do some reading of previous discussions: Why is it not implemented yet? What is the problem? Why do we not get any data right now? Speak to mVolz for further information.

Implementation options

• Research: Can we use an existing external service to implement pdf support in citoid?
• What metadata can we get extracted by a third party service?
• Zotero could be a good candidate, see PR
  • When we submit a PDF Link,, what do we get as a formatted result if Zotero is integrated with Citoid?
• Think about fallback: What would be our fallback if the service is not available or we can't retrieve any metadata
• Maybe just a weblink to the PDF? Read the metadata from the PDF and be done?

Research

• Where does the metadata exist? In the PDF / through the site/service hosting the PDF?
• if both, which of both is more reliable to return complete metadata
• What type of metadata is returned? Does the metadata map to the cite template fields.