Template talk:Find medical sources

Side effects
If accepted, this template may imply some changes in the wording or usage of Template:Reliable sources for medical articles. In addition, it also might be added to Module:Find sources/templates, with the code here becoming a wrapper, similar to the way find sources is now. Mathglot (talk) 01:38, 10 August 2021 (UTC)

Sticky Cochrane query
I'm not getting the right results on the Cochrane link. After getting good first results with myelodysplastic syndrome, I tried clicking on the Cochrane link from other queries at Special:ExpandTemplates for completely differen topics, but every Cochrane search comes up "myelodysplastic syndrome", no matter what the query topic is. Not sure if it's a browser cache issue, or what's gong on yet. (All the non-Cochrane links work as expected after a query-topic change.) Mathglot (talk) 01:49, 10 August 2021 (UTC)
 * , if you can figure out what's going on here, that'd be great, because I haven't so far. Mathglot (talk) 05:31, 10 August 2021 (UTC)
 * Hmm, that's odd. Every instance in the template code where "myelodysplastic syndrome" is mentioned, it's done as, so it shouldn't leave the example situation. &#123;{u&#124;  Sdkb  }&#125;  talk 05:36, 10 August 2021 (UTC)
 * I managed to fix it, although it's hard to reverse engineer the numerous attributes in the search url. The fix involved dropping . I'm not entirely sure why that works, but it does. Now the link seems to execute the correct search, and it isn't sticky on the original search; it executes whatever search the query terms call for from arg1. There's one downside, which is that the link as currently constituted when clicked doesn't populate the search form with the selected keywords, although it does generate the correct results, but that seems quite minor so I think we can declare this solved. ✅. Mathglot (talk) 20:03, 11 August 2021 (UTC)
 * Well, I thought it was, but there are still problems with it. Will have to look into it some more. Mathglot (talk) 17:00, 12 August 2021 (UTC)

Pubmed
Since Pubmed allows filtering by source type it might be a good idea to limit the search results to secondary sources. AFAICT this can be done easily by sticking &filter=pubt.meta-analysis&filter=pubt.review&filter=pubt.systematicreview onto the end of the URL. Spicy (talk) 02:08, 10 August 2021 (UTC)
 * , thanks for the tip, ✅! Mathglot (talk) 02:44, 10 August 2021 (UTC)

The Wikipedia Library
@Mathglot, I imagine that this will mostly be used by experienced editors, who will qualify for an account at The Wikipedia Library. Maybe there should be a link to that resource? WhatamIdoing (talk) 19:57, 10 August 2021 (UTC)
 * , very possibly. In the initial version, I drew a distinction between sources where we could get to a list of sources with one click, such as the ones that are there now, and those which required either additional input, or more than one click. This was the reason I left off Proquest, ICD-10, and ICD-11, and also TWL. (Note that there's an anomaly with the "Cochrane" link, which may put it in that "two-click" category.) I think what to add (or remove) should be up to consensus, and I'll be happy to implement whatever the consensus of the WP:MED group dictates. So, if you can attract more users to this discussion and establish support for it, that would be great. Thanks for your comment. Mathglot (talk) 10:42, 11 August 2021 (UTC)
 * I don't think that ICD-10 or ICD-11 would be useful. You can find the codes there, but the codes won't really let you write a paragraph.  TWL includes ProQuest and also a couple of medicine-focused sources.  I believe there's a way to make text visible only to people with certain user rights (e.g., extended confirmed), so the link could be hidden from people who likely don't qualify for it.
 * Astinson (WMF), is there a way/URL that would make https://wikipedialibrary.wmflabs.org/users/my_library/ filter for any of several tags? I'd love to give people a link that takes them straight to a pre-filtered list of relevant resources at TWL (e.g., life sciences+health+multidisciplinary). WhatamIdoing (talk) 15:02, 11 August 2021 (UTC)
 * , Thanks for the comment; so no ICD links then. Whether ProQuest is in TWL doesn't affect whether we should include it here, as ProQuest is accessible to anyone, regardless whether or not you're enrolled in TWL, e.g:
 * Find sources . . . ProQuest
 * The reason it's not there now, is that it gives you the search box and you have to enter the keywords and hit Enter, whereas with all the other links, you don't because the links go directly to the sources. But maybe it's worth adding at the end, since it has lots of resources. And yes, one could use If EC to test for extended confirmed.
 * One thing I've been wondering about, is if we could associate a public library id with a user (securely, so in Preferences, perhaps) then a lot of solutions become available for users logged in to their library, such as exproxy, EBSCO, ProQuest, and many others. Mathglot (talk) 08:30, 12 August 2021 (UTC)
 * Surprisingly, it seems that ProQuest gives me different/non-overlapping results depending on whether I'm logged in or not. I only tried one search term, and it was fairly obscure (Pseudoisochromatic plate).  When I was logged in via TWL, I got one (excellent) book result during the last five years.  When I was logged out, I got 114 results – but not that one.  I don't know what to make of this. WhatamIdoing (talk) 23:22, 12 August 2021 (UTC)
 * Database consolidators like JSTOR, ProQuest, EBSCO, etc. have access to dozens of individual sources that they query, before compiling the result set from all of them into one list of results for you. While I don't know what's going on under the hood in your case, it wouldn't be unusual for them to offer more databases to logged in users; so that perhaps, your logged-out search came from 11 databases, and your logged-in one from 37; thus accounting for the differences. Just a guess. Could also be that logged-in/logged-out gives you different sets of default features, like reviews only, digital collection only, free-to-me only, and so on. Or, it could be moths. Mathglot (talk) 23:46, 13 August 2021 (UTC)
 * What's surprising to me is that I got fewer sources while logged in. WhatamIdoing (talk) 23:05, 14 August 2021 (UTC)
 * Hm, that does seem strange. Mathglot (talk) 08:41, 15 August 2021 (UTC)
 * Regarding tags - you can only filter for one tag at a time, but you can add that to the URL, like so: https://wikipedialibrary.wmflabs.org/users/my_library/?tags=health-sciences_tag. This link will also include multidisciplinary resources soon (T289253). In terms of ProQuest, I think it's actually probably the opposite to the situation Mathglot suggested - namely that when you search on ProQuest.com without going through the library, you're searching all their content. When you go via TWL you're being limited to the databases which we have access to as part of our agreement, which is a subset of the overall content. Samwalton9 (WMF) (talk) 16:10, 2 September 2021 (UTC)
 * Thanks, @Samwalton9 (WMF). I'm looking forward to having the multidisciplinary category added to all the others. WhatamIdoing (talk) 01:08, 3 September 2021 (UTC)

Conversion to Module:Find sources
Once this stabilizes, it should be converted to be subsumed under Module:Find sources, starting with the creation of Module:Find sources/templates/Find medical sources based on the links in this template, which would be converted to transcluding the module, analogous to Template:Find video game sources. Mathglot (talk) 17:50, 12 August 2021 (UTC)


 * This is currently being discussed at the Village Pump.
 * Thanks, Mathglot (talk) 23:54, 13 August 2021 (UTC)
 * Thanks, Mathglot (talk) 23:54, 13 August 2021 (UTC)

CiteSeerX
It appears that CiteSeerX stopped indexing medical journal content in 2017. Any objection to removing and perhaps replacing with Credo or PMC? Credo is nice because there's always at least one free encyclopedia entry for each topic. The company is associated with so many libraries that it's easy to get access to all results. - Wikmoz (talk) 22:24, 25 September 2021 (UTC)
 * You must be right; I typed "covid" into CiteSeerX and got, "a system for collaborative virtual 3D design". Oops!
 * I tried Credo on my "medium-rare query" and got nothing (here), so maybe not a great choice, at least, for that type of query. For PMC, the link I found for it is &mdash;is that the one you meant? That link has great results for my query. But there's a PubMed link already in the current template, at the slightly different address , which gives different results. I don't think we should have two PubMed links in the list, but which of those is the right one?  As far as adding a link to replace CiteSeerX, do you have a third suggestion? Or, do you think Credo is worth a second chance, maybe for easier searches? Mathglot (talk) 00:35, 26 September 2021 (UTC)
 * PubMed Central (PMC) is a repository of open access journal articles. So while PubMed returns both free, abstract, and citation-only results, PubMed Central only returns the free stuff. I pictured a (PMC) link in parenthesis after PubMed but I guess users could accomplish the same end by using the "Free full text" filter on PubMed (or we could set this by default with the "&filter=simsearch2.ffrft" search parameter). I guess it depends on the percentage of users that we think are going to have institutional access. I see what you're saying about limited results for long-tail queries in Credo. I initially thought of it as a good replacement for Infoplease in Template:Find biographical sources. It might work better there. Perhaps we can leave the CiteSeerX slot empty until something else comes up. I'll try to think of some other resources. - Wikmoz (talk) 00:59, 26 September 2021 (UTC)
 * , I think Trip Database may be the best option. Added for testing but feel free to remove if you have any concerns. - Wikmoz (talk) 20:02, 19 October 2021 (UTC)
 * Tried the same query and got (these results); looks like a good addition. I'm not so familiar with most of these dbs, and one thing we should keep aware of is independence of results, so if Trip (or any of them) pull from the same sources as some other db, we should consider keeping just the one with better results. I don't suspect that here, just throwing it out there so I don't forget. Mathglot (talk) 20:58, 19 October 2021 (UTC)