Sonny Software

I have 1,178 records with attachments. Unfortunately, Bookends is not able to automatically retrieve metadata for these records. What is the recommended approach to tackling this problem?

I can think of one clunky solution: move these attachments into another application (e.g., Zotero). And hopefully, Zotero is able to auto-detect. Then, if possible, bring back (import) those attachments with the metadata into Bookends. I'm not sure if this is even feasible.

I would be grateful for any suggestions.

It is clunky, but feasible. Autocompete Paper (in Bookends) is designed for instances like this, but it's meant for a few PDFs, not thousands. You might ask why very few PDFs were not found. Do they have DOI's? If so, know that Crossref (which Bookends uses to fetch metadata based on DOIs) has been very slow recently and references haven't been fetched as a consequence. I've reported this to them, with no answer so far. But I'm sure they'll fix it, and if that's the problem you're having Autofill From Internet (which can work in batch) will work again when they do.

Jon
Sonny Software

Thanks, Jon. I will try the Zotero route; however, if there are other suggestions, I'd love to hear them.

The Origin Story of my anonymous references is a long (and sordid) tale (I don't think Marvel is interested). My issue is probably not due to a recent slow-down or glitch in Crossref. I am embarassed to say that the 1,178 records have been sitting in my Bookends database for years. They spawned from a move from another reference manager. The sheer size was de-motivating. I dealt with them when the need arose. When I go on a hunt for a PDF --- usually an urgent affair brought on by a deadline --- I do not come back to address the root problem. My conscience knows the fragility and foolishness of this method, if one can call it that.

The glass is half-full: The 1,178 records stand against 2,175 records with proper metadata. Two-thirds are in good standing. Now, I have to admit that of the 2,175 records many needed manually "nudging" to get right.

Below are anecdotal observations of where auto-detection seems to fail when I feel it shouldn't:

1. the PDF contains the DOI, but Bookends does not seem to detect it in the PDF (see 2a below). Something off about how Bookends scrapes the PDF in these cases? Don't know and haven't examined closely.

2. the PDF has more than one: once to supplemental data, and another that refers to the actual paper. Specific example: "`www.pnas.org/lookup/suppl/ doi:10.1073/pnas.1105901108/-/DCSupplemental`" and "`www.pnas.org/cgi/doi/10.1073/pnas.1105901108`". This throws off Bookends, if it picks up the DOI for the supplemental first.

2a. In fact, I think (not sure) it throws off Bookends if the DOI is part of a URL.

Finally, I have a feature suggestion to assist auto-detection. I'll put that into a separate thread.

Bookends will find DOIs if they are in a URL AND if they terminate the URL. If they're in the middle Bookends won't find the correct one.

Have you tried Autofill From Internet on, say, 100 at a time?

Jon
Sonny Software