Improvements to the Category Walker

Any posts related to the categorization and standardization of IMSLP

Moderators: vinteuil, Davydov

KGill
Copyright Reviewer
Posts: 1295
Joined: Thu Apr 09, 2009 10:16 pm
notabot: 42
notabot2: Human

Re: Improvements to the Category Walker

Post by KGill »

imslp wrote:This may have been brought up before, but would it be possible to consider the tags to actually provide certain information about the piece, like the Composer and Piece Style? That way we can put the links at the end of the GI box with the heading "Genres" or "Tags".
This is almost exactly what I was thinking of :) I guess I wasn't making myself all that clear. (Sounded clear in my head... :roll: ) Do you think it might be possible to make the placement of the links more specific to the attribute that they're describing? For instance, the Instrumentation field could link to the instrumentation category, the language field to the language category, etc. (This is where I was going with my earlier question to you about differentiating between types of tags.)
imslp
Site Admin
Posts: 1642
Joined: Thu Jan 01, 1970 12:00 am

Re: Improvements to the Category Walker

Post by imslp »

Not yet, as MW:G does not formally distinguish between the different types.

But I think it will certainly be possible in the future to implement the system such that Steltz's concerns are also addressed, e.g. that the categories may be properly sorted and become (more of) a source of information about the piece, especially considering that we've phased out the old genre system (which used to occupy a similar position in the GI box).
steltz
active poster
Posts: 1861
Joined: Sat Dec 13, 2008 2:30 pm
notabot: 42
notabot2: Human

Re: Improvements to the Category Walker

Post by steltz »

Now that I've seen the changes, I just have to say that in my opinion, the flow of the page is not ideal. It now goes from the specific to the general and back to the specific. The Instrumentation field at least needs to go above the genre categories. In pages with many many categories, the instrumentation will be even further separated from the other specific information.

Can we at least do all the specific information first?
bsteltz
imslp
Site Admin
Posts: 1642
Joined: Thu Jan 01, 1970 12:00 am

Re: Improvements to the Category Walker

Post by imslp »

Actually, I did exactly that last night. It may take a while for cache to clear though.
steltz
active poster
Posts: 1861
Joined: Sat Dec 13, 2008 2:30 pm
notabot: 42
notabot2: Human

Re: Improvements to the Category Walker

Post by steltz »

Thanks -- it must be the "corrector of graduate research papers" in me -- flow and logical ordering and all . . . .
bsteltz
haydenmuhl
active poster
Posts: 203
Joined: Thu Oct 28, 2010 6:20 pm
notabot: 42
notabot2: Human

Re: Improvements to the Category Walker

Post by haydenmuhl »

I'd like to raise an issue I've encountered with the genre walker.

The genres are attached to the work pages, as opposed to the scores themselves. That means that if I want to find a piece that is voice, piano and violin, I may get false positives. For example, when I perform the search "Intersect: Scores featuring the violin AND For voice, piano", I get Winterreise by Schubert, because it is a cycle for voice and piano, and there is a violin transcription of one of the songs. That search would intuitively mean to me, "Give me pages that contain scores for exactly one voice, one piano and one violin."

I realize that any fix would be a completely new implementation from the existing tagging system, but I wanted to just voice the idea.
Classical Voices - a forum for classical singers
pml
Copyright Reviewer
Posts: 1219
Joined: Fri Mar 16, 2007 3:42 am
notabot: 42
notabot2: Human
Location: Melbourne, Australia
Contact:

Re: Improvements to the Category Walker

Post by pml »

The subtlety here is that a category of the form “scores featuring P and Q” usually refers to subsets of instruments found in works (or their arrangements), whereas the formula stating “for X, Y, and Z” is usually a complete (if sometimes abbreviated) specification of the instrumentation of the works in question. So if you really want works for violin, voice, and piano (all 3) you need to specify that category more exactly: try walking from the starting point of “for voices with solo instruments” on the main Work genre page under Instrumentation; you will find the second page shows 11 works “for voice, violin, piano”. (If there had been a few more examples of this work type, the category you wanted would have been listed in the first 100 results on the initial page.)

Tricky!
--
PML (talk)
steltz
active poster
Posts: 1861
Joined: Sat Dec 13, 2008 2:30 pm
notabot: 42
notabot2: Human

Re: Improvements to the Category Walker

Post by steltz »

You can also approach it from the "for 3 players" category.
bsteltz
pml
Copyright Reviewer
Posts: 1219
Joined: Fri Mar 16, 2007 3:42 am
notabot: 42
notabot2: Human
Location: Melbourne, Australia
Contact:

Re: Improvements to the Category Walker

Post by pml »

Not so fast Steltz! The voice isn’t counted as a “player” ;-)
--
PML (talk)
steltz
active poster
Posts: 1861
Joined: Sat Dec 13, 2008 2:30 pm
notabot: 42
notabot2: Human

Re: Improvements to the Category Walker

Post by steltz »

I guess I haven't looked for vocal works yet!!! I assumed "players" meant any of the musicians. Ignore my advice. :oops:
bsteltz
pml
Copyright Reviewer
Posts: 1219
Joined: Fri Mar 16, 2007 3:42 am
notabot: 42
notabot2: Human
Location: Melbourne, Australia
Contact:

Re: Improvements to the Category Walker

Post by pml »

For what it’s worth, my initial reaction was identical to yours, why not try walking the “for 3 players” category? However I checked and soon found it wasn’t the right way to attack the problem.

The Work genre page is actually quite clear at separating out purely instrumental categories from vocal/instrumental ones — by means of a nice dividing line.

By the way, starting from voice with solo instruments shows a much larger number of scores for voice, violin, and continuo, if you’re willing to accept the anachronistic use of a piano as a continuo instrument.

Cheers, PML
--
PML (talk)
haydenmuhl
active poster
Posts: 203
Joined: Thu Oct 28, 2010 6:20 pm
notabot: 42
notabot2: Human

Re: Improvements to the Category Walker

Post by haydenmuhl »

pml wrote:The subtlety here is that a category of the form “scores featuring P and Q” usually refers to subsets of instruments found in works (or their arrangements), whereas the formula stating “for X, Y, and Z” is usually a complete (if sometimes abbreviated) specification of the instrumentation of the works in question.
I wasn't aware of this. This will help how I search in the future, but it doesn't actually address the issue I brought up.

For example, "Scores featuring the voice AND Scores featuring the clarinet" (link). I would want this to find me scores that include both a solo voice and a solo clarinet, possibly plus other instruments. I would expect to find "Hirt auf dem Felsen" in the search results, and indeed I do.

However, the Rachmaninoff vocalise (14 Romances) is also included, and probably shouldn't be. It is included because 14 Romances is a work page that includes a score featuring the voice and a score featuring the clarinet. There is no score on that work page with parts for both the voice and clarinet. If these tags were attached to scores, instead of works, the search would work as expected.

Again, the search is currently useful, even with this behavior, and I realize it would take a lot of work to correct this.
Classical Voices - a forum for classical singers
pml
Copyright Reviewer
Posts: 1219
Joined: Fri Mar 16, 2007 3:42 am
notabot: 42
notabot2: Human
Location: Melbourne, Australia
Contact:

Re: Improvements to the Category Walker

Post by pml »

Again, you’re confusing matching overlapping subsets with the precise specification. Yes the use of the AND operator does imply an exclusive intersection of the two categories – except that an individual work can have multiple different arrangements.

As I said, to get a precise instrumental grouping you should go to the main View Genres page, select at “for voices with solo instruments” (exactly the same as the previous example) and browse from there, since that description equally well fits works like the Schubert works with obbligato instruments such as the clarinet in Der Hirt auf dem Felsen and the horn in Auf dem Strom.

Adding a whole lot of “Scores featuring X”, “Y”, and “Z” together in the way you’re suggesting, in the hope of obtaining only the scores which actually matches those qualifications will not work until every single score on the website is individually tagged – which is a task that represents an order of magnitude more work for the library staff. (Hint: it’s probably not going to happen any time in the foreseeable future.)

Cheers, Philip
--
PML (talk)
haydenmuhl
active poster
Posts: 203
Joined: Thu Oct 28, 2010 6:20 pm
notabot: 42
notabot2: Human

Re: Improvements to the Category Walker

Post by haydenmuhl »

pml wrote:As I said, to get a precise instrumental grouping you should go to the main View Genres page, select at “for voices with solo instruments” (exactly the same as the previous example) and browse from there, since that description equally well fits works like the Schubert works with obbligato instruments such as the clarinet in Der Hirt auf dem Felsen and the horn in Auf dem Strom.
That's fine, but it's similarly broken given my use case. If I am putting together a concert with my clarinetist friend, Auf dem Strom is going to be just as irrelevant a search result as Vocalise.

Looking at actual numbers, there are about 30,000 works and about 90,000 scores. It seems to me the workload of tagging scores would be about three times as much as tagging scores, not necessarily an order of magnitude. Perhaps slightly less, because the 90,000 number accounts for things like orchestral parts. For example, the full score plus parts for a Beethoven symphony make up about twelve scores, but they could be tagged once as one logical score since they all belong together. Then the four hand piano transcription, which is one score, would get its own tags.
Classical Voices - a forum for classical singers
steltz
active poster
Posts: 1861
Joined: Sat Dec 13, 2008 2:30 pm
notabot: 42
notabot2: Human

Re: Improvements to the Category Walker

Post by steltz »

I see the logic in your argument, but as part of the tagging team, I have to point out:

1) if +-35,000 works have already been tagged, but there are 92,000+ scores on the system, then 35,000 tags would have to be moved and 57,000 tags would have to be created. I find it hard to keep up as it is, and every time we get the 2,700 untagged pages down to about 2,600, somebody finds a new collection of a couple of hundred songs (not that I'm complaining :D ), uploads it, and our backlog that we thought we reduced is right back up there. That might sound like an exaggeration, but our milestone counter is clicking off another thousand every couple of weeks -- and the last thousand came in 10 days, which is, in fact, a couple hundred new files every 2 days.

2) When I checked out the "for voices with solo instruments" page (since I obviously was a little ignorant about it), it took me less than a minute to locate about 16 groupings that would accommodate voice and clarinet.

I don't want to shoot down what sounds like a good idea, but I don't see that it's worth many hours that we don't have, when there's another way to do it that doesn't really take too long.
bsteltz
Post Reply