Automatic tags

rlwither

What are all of the sources of automatic tags when they are drawn from an RIS record? Specifically, if there are no assigned subject terms or keywords, where else might automatic tags be drawn from?


* I'm an academic librarian and have been in charge of supporting Zotero at my university for several years. I am working with 2 other librarians and a researcher to get the answer to this for a research project.

adamsmith

I don't quite understand the question. You seem to be alluding to unexpected behavior somewhere, if that's the case, could you describe that?

Imports from RIS doesn't create automatic tags at all. They only get the tags in the RIS file, and those aren't marked as automatic when importing the RIS directly.

rlwither

It's not so much unexpected behaviour.  We're just trying to understand where in the record the tags are drawn from.  Are they author keywords?  Subject headings?

rlwither

Other places?

adamsmith

That really depends on the RIS -- there are no rules for that at all. It's just the KW tag in the file and you'll see huge variation depending on where you get it, including all of the things you list above plus other keyword taxonomies such as JEL keywords from some publishers.

Same answer if you're not specifically asking about RIS (are you?) but about Zotero import in general: Zotero imports as tags what it things the best available keywords are. That's e.g. typically (LoC) subject headings from library catalogs, MeSH terms from Pubmed, and a hodgepodge of things from other sources.

juliemorris

Hi, rlwither is helping me find a solution.


I uploaded an RIS file that was exported from JSTOR. There were no ‘keyword’, ‘subject’, ‘classification’ or other similar fields in the RIS file that I uploaded. So, we’re wondering how Zotero came up with those automatic tags if they’re not provided within the dataset? Does Zotero use natural language processing to identify keywords/tags (in titles, abstracts, etc.)?

dstillman

@juliemorris: adamsmith answered this above.

There were no ‘keyword’, ‘subject’, ‘classification’ or other similar fields in the RIS file that I uploaded.

I'm not sure what this is supposed to mean. Those aren't RIS fields. The keywords are in the KW tag in the RIS file. You can see them if you open the file in a text editor (e.g., Notepad or TextEdit).

adamsmith

That's why I was asking about unexpected behavior above. Could you give us precise steps to reproduce this? Exactly which file are you exporting, how are you importing it, where are you checking for tags in Zotero.

Given what I know about JSTOR (which doesn't include keywords in its RIS), what you're describing should not be possible -- Zotero will definitely never use anything other than existing metadata to generate keywords, so something else is going on.

juliemorris

@dstillman

Thank you SO MUCH. I have not worked with RIS files in the past - I wasn't aware that pulling it into Notepad would reveal all the underlying metadata. This is exactly the answer I was looking for.

This resolves my issue!