Inspector: Searching Full Text content#387

Hi, yet again! I just wanted to clarify that you are only searching for words in the title and abstract when you do a search in NK or PubMed. Am I correct that the full text is not being searched? I wanted to clarify this for the wiki.

The full text is not being searched!

It is just tiab

Good to know, Nicole. I am assuming that this is going to be true for a long time into the future. So I will make that clear in the wiki!

On NK inspector: No, we only currently offer tiab searching. Maybe at some point we could offer PDF searching, that would be cool!

On PubMed: This is out of our control, as we’re just passing along your query to PubMed. But the general answer is no, PubMed doesn’t index using the FT. More here.

I’m repurposing this issue for full-text keyword searching, which is an interesting idea! Not sure other comparable softwares offer this.

Good to know, @Karl Holub.

@Alex Mebane , this is the request I was referring to–I think we all want full-text search in Inspector!

Yep, would love to see a full-text search feature from Inspector. Thanks!

This is a nice value add IMO, and something I’m not aware of in similar softwares; would love to add. It will likely have unideal runtime implications (either on FT upload building an index, or when running your filters). Bonus points it will be fun for developers :) We’ll take a look.

Added in release 1.40.2. Some usage notes:

• Searches are case insensitive, string matches
• We are simplifying whitespace (multi-spaces, line breaks, etc.) to a single space to make keyword searching reliable. Just type in phrases with single spaces between words

Caveats to consider:

• Not all PDFs (namely image scans) will be searchable
• Hyphenations are not removed, so some words (as you read them) will not be properly searchable
• You may get some false positive matches due to disparate text sections being combined (e.g. the abstract or author list is directly abutted to the body text for searching purposes).
• Very temporary: we’re currently backfilling our search index for all previously uploaded full texts. This may require until mid-morning (US central time) to complete. After this, the index will be automatically populated for each full text upon import.

Thanks for this idea, excited to see how it’s used & what we can add in the future (e.g. PICO, study design extraction on full texts).