Wikipedia talk:Reference database

Support this but is it technically possible?

I'd support a proposal like this, but is it technically possible? The current Wikipedia servers too often strain under the current load and I wonder if a system like this would push them to the breaking point? Still, if the technical issues can be resolved, I'd wholeheartedly support such a system. I especially like the possibility of having a standardized reference system with a central means of accessing the citations.--Alabamaboy 03:08, 16 February 2007 (UTC)

I think it should be feasible considering that a cached representation of the bioblographic output could be stored; that would effectively reduce the amount of output cached across Wikipedia. Right now the server load for inlining images is apparently acceptable, so inlining cached formatted text should be even easier.

The load on servers could come from new Special: pages taking advantage of the database. For example, "find all articles citing this reference" could create a load problem. If so, those functions could use periodic images of precomputed data (i.e. periodic updates) to avoid using compute time. I do think any technical implementation hurdles are surmountable here. Thank you so much for your comment. Heathhunnicutt 04:24, 16 February 2007 (UTC)

What about integrity of use

While this is an interesting approach, I'd have concerns that once a reference is listed it would be taken as authorative across all articles. How would it be possible to distinguish where information from the source is valid and reliable and where other information has rendered obsolete. Gnangarra 03:24, 16 February 2007 (UTC)

I have the same concern. On the other hand, in theory one could AfD the reference. Also, having the references centralized would allow tracking down each article that cites a questioned reference. So there might be net gain to quality control. Heathhunnicutt 04:21, 16 February 2007 (UTC)

Technical problems

The current cite templates can barely handle many of the citations I use; it's unlikely that a centralized database would be able to, so I would continue to be forced to format references by hand. Here are some issues I have encountered, sorted by difficulty of using the cite templates:

• Papers that have multiple URLs (stable ones from the publisher and free ones from the author).
• Papers from a book (like conference proceedings) that is part of a series.
• Books that have been reprinted by a different publisher than the original. Both the original and reprint details need to be cited.
• A paper that appeared in a journal and later was reprinted in a book. Both the original details and the book details need to be included. Occasionally the original was in a different language and the book includes a translation into English.

I just don't have the energy to pursue getting the cite templates to handle all these; it took enough work to get a series field added to {{cite book}}. In general, citation is more difficult at WP than in journals because in research papers one doesn't cite the old, foundational papers that have since been reprinted, and one doesn't include URLs. CMummert · talk 03:29, 16 February 2007 (UTC)

Thanks for your thoughtful remarks.

1. You seem to be an extreme {{cite}} user and while I salute you I would suggest that we also need to help the entry-level citer.

2. The multiple-URL example you mention is implicitly handled in the example form. No pay-per-view links allowed, so there should be no need for multiples.

3. Part of the reason {{cite}} is difficult to modify may be that it is cumbersome to use and people fear making it more cumbersome.

The others I didn't think needed responding to. Good points, though. Heathhunnicutt 04:08, 16 February 2007 (UTC)

There is generally consensus that if there is no free link to a paper online then it is acceptable to provide a nonfree link, especially when the nonfree link will in fact be free to users at academic institutions. If there is a free link then that is certainly better, but nonfree links in citations are common and certainly not forbidden. Although it is uncommon to see links to the NY Times Select, it is not at all uncommon to see links to SpringerLink. There was some discussion about this recently on the talk page for WP:CITE. CMummert · talk 13:11, 16 February 2007 (UTC)

As a recently novice Cite user, the hardest part was discovering the concept of templates, that there was a full template that could simply be cut'n'pasted, and the unnecessary bits cut out. Switching to a database will NOT make this easier. And I don't want to have to go relearn everything now that I've got a copy of the full template squirrled away. If you really have to have a database, just go scan for all the cite templates in the code. I too, had to wonder how to twist the template to match a publication I knew about. I'm clearly not as good as CMummert, but I'd hate to be forced into some hideous inflexible form. If the main concern is the edit-clutter, then the suggestion below of having the references expandable (but hidden by default when editing except for the +), shows up to me as a good one. 173.206.138.17 (talk) 05:57, 12 December 2011 (UTC)

Contrast with previous proposals

How is this different from m:Wikicat, m:Wikicite, and m:WikiTextrose? (SEWilco 04:05, 16 February 2007 (UTC))

For one thing, I have never heard of those before. One minute...

• WikiTextrose - This one does not seem to be a means of maintaining a repository, this looks like a tool to mine text and objects including citations into a map of who-cites-what-why. I think that would be a neat application of the change I suggest, but that is not what I am suggesting. I am only suggesting a new format for input of the citations. It is true that existing citations would need to be converted, but not as richly as WikiTextrose proposes. My suggestion would make WikiTextrose easier to implement, however.
• Wikicite - Wikicite is in some ways similar to my suggestion. In particular, a central repository of citations is kept. The citation information is kept free-form in Wikicite whereas I am suggesting keeping database information with standard fields and (globally changeable) standard formatting. Wikicite also intends to require a citation for every factual assertion sentence fragment, and I salute them for their ambition!
• Wikicat - When wikicat is complete, my idea will be obsolete and useless. However, any database accumulated in the meanwhile should be easily transferred.

Thank you for bringing all of these projects to our attention, these were great points!

Heathhunnicutt 04:19, 16 February 2007 (UTC)

You might want to look at the sections near the end of Wikipedia:Footnotes. (SEWilco 06:29, 16 February 2007 (UTC))

Vague name

The article should have a more descriptive name than "improvement 1". (SEWilco 17:43, 16 February 2007 (UTC))

True. I was hoping the controvery over WP:WIF would lead to lots of suggestions and wanted to inspire that. Feel free to move it/edit the essay! Heathhunnicutt 20:45, 16 February 2007 (UTC)

Pain to edit articles with inline references

It is a real hassle to copyedit (or structure) articles that have those huge blobs of referencing in them. Same thing really with infoboxes and some other types of templates. Maybe wikipedia needs 3 levels of viewing: reader view, text editing view, and then some sort of popups or expanded view, with all the guts of the refenreces and infoboxes and all. —Preceding unsigned comment added by 72.82.33.250 (talk) 07:13, 20 November 2010 (UTC)