Guest
home
beatmaps
rankings
community
store
help
forum
  1. Forums
  2. Beatmaps
  3. Mapping Discussion
  4. Ranking Criteria
  5. Finalized/Denied Amendments

[Proposal] Standardize Metadata Artist

posted
Total Posts
17
ailv
403 posts
Joined August 2015
Topic Starter
ailv
Currently the metadata RC includes a standardization for the song to be consistent with prior ranked versions:

RC wrote:

If the same song exists in the Ranked section already, the metadata should be followed unless it breaks other rules in the Ranking Criteria or the Official Sources state something completely different.


There does not exist a standardization for Artist names, which should also be standardized across ranked mapsets.

Amending the above rule to:

Updated wrote:

If the same song, or artist exists in the Ranked section already, the respective metadata should be followed, unless it breaks other rules in the Ranking Criteria, or the Official Sources sate something different.


allows for standardization of artists as well. Cases of this can include:

"Diao Ye Zong feat. Kushi/Meramipop" vs "Diao Ye Zong feat. Kushi, Meramipop"
https://osu.ppy.sh/beatmapsets/1013030/discussion#/1123009

and prevent cases such as:

"Ito Kashitaro" vs "ITO KASHITARO"

in the future.
Last edited by ailv , edited 2 times in total.
pishifat
osu! Paragon
3,812 posts
Joined August 2013
pishifat
doesn't "the metadata" in the original quoted rule refer to artist and title fields? i always assumed it did
Noffy
Nomination Assessment Team
1,685 posts
Joined April 2012
Noffy
the proposal means the artist should be consistent across ranked maps of different songs by an artist, whereas the current rule only applies to multiple maps of one song.

the proposal makes sense in concept but could be a bit fuzzy across intentionally differing aliases, i.e. Silent Siren changing to SILENT SIREN when they changed labels. Same goes for capsule vs. CAPSULE ( https://web.archive.org/web/20150614115008/http://www.musicman-net.com/artist/27938.html ). Currently things like this "follow what the song's release says" but standardizing artists across different songs could go against that practice.

so implementing such a guideline could mean potentially changing what's intentionally different, and when it comes to messing around with artist names directly that's a bit of a sensitive matter. It could also potentially be more work to try to figure out "ok did this person change their artist name intentionally (so it should be different from other song's maps), or did they do it by mistake (so it should be changed to be the same as other maps of songs by the artist)"


I get that the proposal is for when there's multiple artist choices across multiple songs by them and people choose inconsistently, but I'm struggling to imagine how to implement such a guideline without effecting things like I mentioned above.

tl;dr : good in concept, can't imagine how to implement as a guideline that clearly covers different case types
ailv
403 posts
Joined August 2015
Topic Starter
ailv

Noffy wrote:

the proposal means the artist should be consistent across ranked maps of different songs by an artist, whereas the current rule only applies to multiple maps of one song.

the proposal makes sense in concept but could be a bit fuzzy across intentionally differing aliases, i.e. Silent Siren changing to SILENT SIREN when they changed labels. Same goes for capsule vs. CAPSULE ( https://web.archive.org/web/20150614115008/http://www.musicman-net.com/artist/27938.html ). Currently things like this "follow what the song's release says" but standardizing artists across different songs could go against that practice.

so implementing such a guideline could mean potentially changing what's intentionally different, and when it comes to messing around with artist names directly that's a bit of a sensitive matter. It could also potentially be more work to try to figure out "ok did this person change their artist name intentionally (so it should be different from other song's maps), or did they do it by mistake (so it should be changed to be the same as other maps of songs by the artist)"


I get that the proposal is for when there's multiple artist choices across multiple songs by them and people choose inconsistently, but I'm struggling to imagine how to implement such a guideline without effecting things like I mentioned above.

tl;dr : good in concept, can't imagine how to implement as a guideline that clearly covers different case types


We already face the issue of not knowing whether a new stylization choice is intentional, or mistakenly, and note that this only effects stylization, if an artist chooses a new name to release there songs as, it'd be clear that that song should be using that artist for it's metadata, and can be considered, in some ways as a separate artist.

Notably as a guideline, this means it can still be handled on a case-by-case basis, and so leeway can be afforded. There really shouldn't be a reason that two songs on the same album have differing artists, as is the case with https://osu.ppy.sh/s/875648 and https://osu.ppy.sh/s/886458, which can cause confusion.
ailv
403 posts
Joined August 2015
Topic Starter
ailv
Another recent case of this would be https://osu.ppy.sh/beatmapsets/833772#osu/1746520

osu.ppy.sh/beatmapsets/1001507/discussion/-/generalAll#/1159270,

Where it doesn't make sense to have them split like that, with another set in qualified as well https://osu.ppy.sh/beatmapsets/921763#osu/2042325 that's from the same album.

A potential solution here would be to standardize metadata across all ranked songs from the same album, but leaving case-by-case basis by having it as a guideline would still work better.
Noffy
Nomination Assessment Team
1,685 posts
Joined April 2012
Noffy
same album would be finnicky because of songs appearing on different albums and requiring a different level of meta checking compared to normal

idk wrote:

Artist names and field formatting should be consistent between different songs from the same person or group. This is unless the person or group intentionally uses a different alias for different song or album releases


?
idk if that's clear enough but was best I could think of. Can probably be improved (?)
Last edited by Noffy , edited 2 times in total.
pishifat
osu! Paragon
3,812 posts
Joined August 2013
pishifat

Noffy wrote:

Artist names and field formatting

what's the difference? cant it just be "artist field formatting"?
Noffy
Nomination Assessment Team
1,685 posts
Joined April 2012
Noffy
mainly for artists that show up in different collabs and stuff, consistent formatting wouldn't cover that.

Like Codama Saori should be romanised that way even she is in some collab that's like

other anime singer, Codama Saori

where formatting isn't same as just "Codama Saori", but the artist name is made consistent


consistent formatting would mean

you can't have 30 maps formatted
Dyz feat. Singer
then decide 1 map would be just Dyz and not mention feat. Singer
since that'd break consistency


edit: Gonna give this a week for discussion please discuss (please)
deadline for discussing October 3
Last edited by Noffy , edited 4 times in total.
pishifat
osu! Paragon
3,812 posts
Joined August 2013
pishifat
im ok with this
Serizawa Haruki
1,696 posts
Joined June 2015
Serizawa Haruki
I support this proposal, however the wording only includes intentionally using a different alias. As already mentioned, it's not always clear whether something is intentional or not and also, how would this be handled if an artist is labeled inconsistently on different songs/albums/websites by mistake? In those cases you would probably need to add that the most common version should be used.
For example, I noticed that Reol is sometimes written in caps and sometimes not (https://osu.ppy.sh/p/beatmaplist?l=1&r=0&q=reol&g=0&la=0&ra=&s=4&o=&r=0&q=reol), it would be nice to use one of the two and stick with it (unless there is a specific reason why some songs use REOL and some Reol).

The same goes for unintentionally inconsistent formatting, let's say an album includes the following tracks:
Artist1 x Artist2 - Song1 (feat. Singer)
Aritst1, Aritst2 - Song2 feat. Singer
Artist1 & Artist2 feat. Singer - Song3

This is obviously an exaggerated example but it's very well possible that the songs were uploaded to YouTube or something and that the format for the artists is different there
Last edited by Serizawa Haruki , edited 7 times in total.
ailv
403 posts
Joined August 2015
Topic Starter
ailv

Serizawa Haruki wrote:

I support this proposal, however the wording only includes intentionally using a different alias. As already mentioned, it's not always clear whether something is intentional or not and also, how would this be handled if an artist is labeled inconsistently on different songs/albums/websites by mistake? In those cases you would probably need to add that the most common version should be used.
For example, I noticed that Reol is sometimes written in caps and sometimes not (https://osu.ppy.sh/p/beatmaplist?l=1&r=0&q=reol&g=0&la=0&ra=&s=4&o=&r=0&q=reol), it would be nice to use one of the two and stick with it (unless there is a specific reason why some songs use REOL and some Reol).


REOL is the 3 man unit that disbanded consisting of Reol, GigaP, and Okiku, which accounts for most of the differences (there's 2 errors i see).

Serizawa Haruki wrote:

The same goes for unintentionally inconsistent formatting, let's say an album includes the following tracks:
Artist1 x Artist2 - Song1 (feat. Singer)
Aritst1, Aritst2 - Song2 feat. Singer
Artist1 & Artist2 feat. Singer - Song3

This is obviously an exaggerated example but it's very well possible that the songs were uploaded to YouTube or something and that the format for the artists is different there


Note that the entire point of this proposal is to serve as a solution for these "What Ifs?" since the point is to standardize them.
tatatat
923 posts
Joined March 2015
tatatat
I assumed the song, artist, and source should all match previous ranked maps. If this is not the case, it should be.
Serizawa Haruki
1,696 posts
Joined June 2015
Serizawa Haruki
I probably didn't explain my point properly. While the guideline says that the artist name should be standardized across different songs from the same artist, it does not specify which "version" of an artist name should be used for future maps if there are multiple ones. In order to sort this out, I think it would be useful to add something along the lines of
"If an artist is labeled inconsistently in different official sources, use what is most common."
This is for cases like Ito Kashitaro as mentioned in the OP.

On a side note, I noticed something odd while reading the RC: There is no rule that says you have to use the capitalization of officially romanized artist names. For example, mafumafu (まふまふ) is always written lowercase by himself in official sources. However, if you apply the hepburn romanization method and use standard capitalization, you would get Mafumafu. Technically you don't have to use his official romanization since there is only an allowance regarding it, not a rule:
"If an artist has provided an official translation for their name, this may be used in the romanised artist field. Official romanisation may be used for the spelling of an artist's name, but the name order must follow the related rule."
Therefore it's actually optional and I don't think that was intended, or am I missing something? The same goes for artists whose official romanization doesn't follow the hepburn method such as KISIDA KYODAN & THE AKEBOSI ROCKETS (岸田教団&THE明星ロケッツ). One could theoretically choose to ignore the allowance in the RC and use "Kishida" and "Akeboshi" instead
Last edited by Serizawa Haruki , edited 1 time in total.
Noffy
Nomination Assessment Team
1,685 posts
Joined April 2012
Noffy

haruki wrote:

"If an artist is labeled inconsistently in different official sources, use what is most common."

I could agree w/ this, but that also requires potentially finding every source to try to determine what's "most common" in some cases. Common or reasonable could work, but could be tricky to enforce.

haruki wrote:

On a side note, I noticed something odd while reading the RC: There is no rule that says you have to use the capitalization of officially romanized artist names.


yes, that's a thing. sometimes artists would have like, you know, 1 romanisation all lowercase on some album for 5 years ago and then not romanise it anywhere else. should it be considered the romanization? in a lot of cases like that, not really. but then requiring official romanisation to be used when keeping this in mind would require defining what sources or amount of frequency are/aren't valid for that. Rating the validity levels of different types of official sources... I don't think that would be beneficial in the long run.


Basically the issue with both these things is requiring people to determine what's less/more valid as a choice than other things, instead of being able to pick 1 allowed official source and 1 allowed romanisation method (hepburn vs. official) and rolling with it (simplest), and then maps after that following suit for consistency.
Last edited by Noffy , edited 1 time in total.
Noffy
Nomination Assessment Team
1,685 posts
Joined April 2012
Noffy
While I do see merit in other points brought up in this thread, i.e. how to decide "most important version" of metadata, I think it is a bit out of scope for what the thread originally proposes. The song guideline has existed for over a year without major issues from the same type of problem, so I do not think it is critical to have those standards in place for similar artist consistency.
If anyone feels it needs further discussion or related rules of their own, feel free to open up additional threads.


As for the original proposal, this is now PR'd on GitHub and will be merged in a few days unless major concerns with the guideline itself appear.
Serizawa Haruki
1,696 posts
Joined June 2015
Serizawa Haruki
I'm fine with it, I was just concerned that it might cause ambiguity or misunderstandings but if you think it will work out then I fully support this change
Noffy
Nomination Assessment Team
1,685 posts
Joined April 2012
Noffy
Oh I see, thanks for replying about that, good to keep in mind ^^

The PR has now been merged!

Edit: In this PR "Ranked" for this guideline and the similar song guideline has been changed to "Ranked or Loved" as Loved has had metadata actively checked for every map for over a year now. While old loved maps have very questionable metadata, or sometimes bad ones get through, this is the same case as for Ranked.
Last edited by Noffy , edited 2 times in total.
Please sign in to reply.

New reply

0
1 / 17