Retrieving UWorld Question IDs from tags in AnKing and Retiring Defunct UWorld QIDs

I’m going to demonstrate step-by-step how to retrieve UWorld Question IDs from tags within AnKing and systematically remove tags for retired UWorld Question IDs.

put in tag; select all

Click “yes”

Click generate test

Test is generated

However, much of the time, there are QIDs generated that in fact do not exist within UWorld, as shown here:

Some, but not all, of these QIDs are tagged by UWorld as not working

Plugging in just those test IDs alone obviously doesn’t work

Removing the tagged QIDs generates a functional test in UWorld

Therefore, we can deduce that the Question IDs in the test that didn’t work have been retired from the Step 1 Qbank, and that their tags in AnKing should therefore be removed.

Those may be comlex numbers… comlex and step have different numbers

1 Like

The tags all say “UWorld” though and the QBank is Step 1 not level 1

These are the tags: tag:#AK_Step1_v12::#UWorld::6575* OR tag:#AK_Step1_v12::#UWorld::6685* OR tag:#AK_Step1_v12::#UWorld::26106* OR tag:#AK_Step1_v12::#UWorld::24628* OR tag:#AK_Step1_v12::#UWorld::8286* OR tag:#AK_Step1_v12::#UWorld::24764* OR tag:#AK_Step1_v12::#UWorld::6678*

Unless the COMLEX tags are labeled as being Step1_v12 tags then I’m pretty sure these are Step 1 tags

Thank you for your detailed write up, if anyone has COMLEX Level 1 UWorld and can check those ID’s to double check that would be great. I’ll check with the maintainers and see if anyone has access currently

3 Likes

Sounds like a good plan. I have a list of all the QIDs that didn’t work (I found 1714 of them). My guess based on what Nick (by the way, thank you for everything you’ve done for medical education) said is that it’s likely some mix of COMLEX IDs and retired QIDs. Someone with a COMLEX account could probably find which ones of the ones I found are actually retired and delete them. In any case, I think it may be a good idea, if there are 2 Qbanks combined under one set of tags, to separate them into two separate groups of tags, one that works on MD UWorld and another that works on DO UWorld. It would be nice for AnKing users to be able to search backwards for UWorld QIDs related to a given B&B video or topics they’d covered in a lecture or first aid without having to delete QIDs that were only found in the COMLEX Qbank.

Anecdotally, I think that an easy method of going backwards from tags to QIDs in this way would make UWorld a much more approachable resource. A big cause of the decline in step 1 pass rates is that students aren’t doing enough UWorld (I’ve seen data from my school to that effect) and a big reason for that imo is that it’s just so big and people don’t know how to approach it and are worried, whether it’s justified or not, about doing problems that won’t help them for their next exam. If people could search backwards they’d be able to find subsets of problems that are most relevant to what they’re studying and that in turn could increase the number of UWorld problems people do.

@Ahmed7 @TheAnKing I have UW COMLEX level 1 - In my experience, QIDs in the 20,000s range are from COMLEX and QIDs in the 100,000 range were new questions being tested.

The best solutions I see that would satisfy @NotEliTanenbaum 's problem/request are:
A) reorganize the anking tag hierarchy to something like #AK_Step1_v11::#UWorld::STEP::[QID] and #AK_Step1_v11::#UWorld::COMLEX::[QID].
B) Edit the UW QID add-on to have both step [#] and complex level [#] in the test dropdown menu. Then have the script exclude the respective QID ranges for the reciprocal test. I’d see this possibly happening by redefining “Relevant Tags” within the script with these conditions or by adding an ‘if not’ function when the script starts to append the proper tag prefix per the selected test/deck version.

For both options, it would be great if there was an official statement from UW on what QID ranges correlate with which product version currently and in the future.

1 Like

UW has refused to work with us… that’s the problem…

2 Likes

Happy to know this is still being discussed. I second what @cpochat is saying. For the 858 QID tags in the google doc I posted (I counted them incorrectly in my previous post and then accidentally doublecounted by doing control f “tag” in the doc) ( Defunct and/or COMLEX tags - Google Docs), I think it’s fair to assume since they don’t populate in MD Step 1 UWorld that they are either retired or are COMLEX IDs, with the latter being the more likely scenario. I think pretty much all the Question IDs that I posted in the doc should have their tags changed to being #AK_Step1_v12::#UWorld::COMLEX::[QID]. I’m not a programmer so I’m not sure how the programming changes in the UWorld addon would work but I think it’d be good to have a non-buggy topic to UWorld QID system and that revamping those tags (and other ones-- I went through most of the UWorld Step 1 tags but not all of them) would go a good step of the way towards getting there.

The COMLEX tags could be cleaned later using the same method I outlined here; if reverse-searched COMLEX QIDs didn’t work in COMLEX UWorld in the new tagging system, then they could be safely retired from AnKing.

@Ahmed7 what do you think? I’d have to have the uworld addon updated

1 Like

Im fine with either way, if we definitely know the ones being changed are COMLEX, and then we switch to a ::COMLEX tag then that’s okay and I can start accepting them once you get the addon configured.

If the community wants this then Im all for it!

1 Like

Let’s confirm them first with someone that has a comlex account. Once confirmed, let me know and I’ll see if we can update the addon

2 Likes

Sounds like a plan! There’s also more QIDs that could be found using the same method that I haven’t located yet.

Awesome. We just need someone with a comlex account to confirm things first. @Ahmed7 do we know of anyone?

I asked before in maintainer chat, dont think anyone currently has one afaik

@NotEliTanenbaum @cpochat know anyone that would have one?

For level 1 mine is active until mid Janurary

2 Likes