Pkp-lib: OJS should support author-less publications

[AhemNason]

This issue describes two problems -- a bug and a feature request.

1. The feature request is that OJS should support publications without a specific author. This leads to bad metadata and downstream services, like Crossref, support author-less objects.
2. The section option "omit authors from table of contents" is not respected and the default theme shows authors anyway. This should be addressed in line with how we handle author-less publications.

Original comment:

In journal section metadata/options, there's a qualifier to omit the author's name from a journal section. This is, by and large, a workaround for journals publishing articles that don't have an "author". Stuff like errata or abstracts collections or any other number of things. These are usually uploaded by managing editors or other journal staff who are also authors of legitimate materials.

All themes, including default, display these author names on the table of contents even if the radial to not display them is selected.

You can find the relevant setting in Journal>Sections>Edit>Omit Author from Journal section...

I'd add that this should also omit the author from citations, the article landing page, and any other associated metadata. This whole feature is a workaround due to not being able to zero out an author field, but the downstream impact of that stored metadata is also potentially problematic.

[asmecher]

@AhemNason, what's your take on authorless submissions in general from a librarian/metadata quality perspective? I'd worry about introducing the potential for downstream breakage with other formats that will require creator metadata.

I don't think "forcing" a single author by e.g. preventing someone from deleting the last author record would be a good UI/UX outcome, so if we deem at least one author to be a requirement, then perhaps make it a pre-publication check so that the submission can't be published unless it meets that condition?

[AhemNason]

Alec, I've been asking for authorless submissions since 2008. It's actually the case that more things downstream accommodate "authorless" than not. _Requiring_ an author leads to way more issues with bad metadata.

For example, if I'm a journal manager named "jim john" and I've been uploading an item to my issues called "contributor" for 10 years, and I have ORCID, I'm sucking in ten years worth of "articles" that I don't want. Not everything in a journal has an author. Errata... for example. We have a journal locally that includes a collection of abstracts from their annual conference in every issue. That collection isn't written by a single author, nor is listing 25 authors sensible.

Crossref allows authorless content. JATS, MODS. Instead of an empty element, there just isn't an author element at all.

Even single author requirement is an overstep. Unless you make it section based. I think having section-based options here like you do for abstracts (for both author and email) would really help a great deal with trash metadata in OJS.

A user should never be required to enter metadata that does not exist.

[AhemNason]

Just a note here too that the "omit section title from the table of contents" is also broken on the default theme.

[asmecher]

Sorry, I could've phrased that in a way that poked fewer bears. Also the author email requirement, when a record is appropriate (https://github.com/pkp/pkp-lib/issues/3561). Actually I think we're well situated to remove these requirements, having gradually moved away from conflating the first author with the corresponding user.

Crossref allows authorless content. JATS, MODS. Instead of an empty element, there just isn't an author element at all.

OK, good, likewise DC. Off the top of your head, do you know about METS, ORCID, Google Scholar?

[AhemNason]

I can look into it but I'd suggest ORCID won't matter if there's no author. Google Scholar might. I can look into those two. And get back to you.

[AhemNason]

So, Google scholar is reporting that it can't index something without an author. But almost always these things wouldn't really benefit from indexing anyway. They are almost always _not_ research articles but other things... like errata or other smaller pieces. But, more annoyingly, they are also reporting that:

If there are widespread cases of papers without authors listed, the indexing system will eventually drop the site.

and...

3 articles in an issue of 10 articles that don't have authors would definitely be a red flag for the indexing system.

For what it's worth, I don't think this is reasonable. I'm not confident that we have numbers on the prevalence of these things. It's probably not as high as 3 articles per issue, but it's hard to say. Either way, bad practice is bad practice! Here's the MODs recommendation on Names and it's more or less where I'd sit:

https://www.loc.gov/standards/mods/userguide/name.html

It's not sexy reading, but the gist is this:

The DLF/Aquifer Implementation Guidelines for Shareable MODS Records recommend the use of at least one element to describe the creator of the intellectual content of the resource, if available.

A name may be linked to a uniform title in the record using the nameTitleGroup attribute. A name may be designated as the citation or "main" entry name using the usage attribute.

and also...

Aggregator information: Aggregators commonly use the field as a target for author or subject searching. Even the simplest interfaces offer an author/creator search. In cases of unknown or anonymous creators of resources, aggregators generally remove values indicating this and rely on institutions' local records to convey this information if necessary.

We don't currently support a type attribute or drop-down for names. If something like _the editors_ is the ask for a fake name, it needs to not be coded as a person. For MODS _requirements_, name is "Recommended if applicable" but not required.

MODS is based on AACR2 Cataloging Standards so... this means that the vast majority of libraries respect an empty author field. And it's not great cataloging practice to make descriptive metadata where it doesn't exist (although, in the interest of the moment, historically they've done some wild things like recommended assuming gender when it wasn't obvious and that sucks).

[NateWr]

Apologies for hijacking your issue, @AhemNason, but I've renamed it to better describe the need for author-less publications as you've described.

I think there's a valid case here for publications without a contributor as OJS understands it now. A couple of questions:

1. In such cases, shouldn't the journal be described as the contributor/author? This could be done with support for corporate names or by supporting author-less publications but automatically assigning them to the journal under-the-hood.

2. Asking as an outsider to this: should these things be published as articles in OJS? Do we need a different model to describe errata, for instance, than just calling it an author-less publication? Does JATS/Crossref really want this kind of content described as an article?

[AhemNason]

_In such cases, shouldn't the journal be described as the contributor/author? This could be done with support for corporate names or by supporting author-less publications but automatically assigning them to the journal under-the-hood._

1. This is what Google Scholar would prefer. It's still making up metadata. It's placeholders for empty fields. Google recommended putting in "the editors", but it's worth asking if you want your citation generation for these saying "the editors" in it. Or, if you actually would cite these at all. I bet if you look in Google Scholar under author "the editors" you'll see a wild amount of stuff.

https://scholar.google.com/scholar?as_q=&as_epq=&as_oq=&as_eq=&as_occt=any&as_sauthors=%22The+Editors%22&as_publication=&as_ylo=&as_yhi=&hl=en&as_sdt=0%2C5

This is problematic. "T. Editors" isn't any one editor. It doesn't tell me anything useful. Especially if I'm not recording the editors for every issue in my issue-level metadata (which we also do not do). A placeholder in this section is symbolic. Symbolic metadata is a bad idea. It also muddies the waters far more than just not having an author attributed because now it matches a million other made-up authors.

_Asking as an outsider to this: should these things be published as articles in OJS? Do we need a different model to describe errata, for instance, than just calling it an author-less publication? Does JATS/Crossref really want this kind of content described as an article?_

1. JATS and Crossref both support authorless in their schema. "Want" isn't really much of an issue when you have over 100 million records. Crossref in particular probably just hopes that user-generated metadata gets better broadly.

If you talk to someone in cataloguing, it's clear that metadata generally shouldn't be assumed, it should be specific. And it should describe what's in the file. For example... if I were to receive an article to upload into OJS and the PDF didn't include author information, I wouldn't go looking for it and have it only appear in the metadata. You want to describe the article/galley as best as you can. You don't want to create _fake_ information out of necessity to get through a submission form.

I'm cautious of the can of worms we open if we decide to _not_ call the publications in OJS "articles". It's true that not all things in OJS are articles. If we open the can of worms, we'd be also opening it to having specific metadata fields for book reviews (everyone mangles book reviews in OJS metadata... it's astonishing). Forms per content type. I'm not _against this_. I think it's probably better than all content having the same form.

Either way, an empty author field for content that has no specific contributor is _more accurate_ than a placeholder.

I do appreciate this being blown out into its own issue but the original bug remains that the section-by-section settings for author or journal section display on the TOC are still broken.

[AhemNason]

I _do very much think_ that this doesn't have to be a journal wide option for all publications in a journal. But it should be ok to have an empty author field via section-by-section settings.

[NateWr]

I do appreciate this being blown out into its own issue but the original bug remains that the section-by-section settings for author or journal section display on the TOC are still broken.

I've kept that bug listed in the original issue description. My interest here is to resolve that _in line with our metadata recommendation_. If we decide that author-less submissions is the way to go, we should remove the option rather than fix it (for example).

I'm cautious of the can of worms we open if we decide to not call the publications in OJS "articles".

I wasn't clear enough in my original question. I'm not proposing that we don't call _all_ publications in OJS articles. My question really comes back to what you said here:

it's worth asking if you want your citation generation for these saying "the editors" in it. Or, if you actually would cite these at all.

If we're talking about material that is fundamentally different from articles in OJS, should we be treating them as articles in OJS? Some things, like editorials, probably should have an attribution -- though that should not be T Editors... maybe the journal name itself, put into a corporate name field? And they probably should at least have the option of going through a form of peer review / editorial sign-off (ie - the submission workflow).

Maybe other things, like errata, shouldn't be published as articles, deposited into Crossref as articles, be given a suggested citation, or be put through the submission workflow? Maybe this content shouldn't be described and treated by OJS as an "article" at all?

Does OJS need an alternate content type for things that aren't a "research output"?

[AhemNason]

Hey Nate, I've been thinking about this a lot and I think it's worth pulling in some answers from hosting services. My gut feeling is _yes_, but my other gut feeling is that it's a great way to complicate a process that already feels too complex for some. I'd like to get some feedback from the hosting support folks. Chiefly, @jmacgreg @amandastevens and @mfelczak . James is on vacation until the 13th but I've emailed it to him to follow up on.

[eocarragain]

I can confirm what @AhemNason says about the "Omit author names for section items from issues' table of contents." section flag being ignored in default and manuscript themes in OJS 3.2.1-1.

@asmecher @NateWr , should this bug be broken out into its own issue - getting this fixed would help hide the bad author metadata at least in the UI.

On the more general issue, I fully agree with @AhemNason : it should not be a requirement to include authors, but should be configurable by journal managers on a section-by-section basis (similar to abstracts etc.).

Thanks
Eoghan

[NateWr]

should this bug be broken out into its own issue

Yes, thanks, can you file that for us @eocarragain?