When should you enter aliases?

From guide:
" An alias is an alternate name for an entity. They typically contain common misspellings or variations of the name and are also used to improve search results. View the alias documentation for more details."

But wouldn’t it be better to correct spelling mistakes and not put the wrong name as an alias?

When the wrong spelling is in a track list, then a user who is trying to search will type in that wrong spelling. Having these alternate spellings in the alias allows search to select the correct answer for them.

2 Likes

Had the incorrect title “Billy Jean” not been included as an alias

would phrase “(medley: Billy Jean / Jeopardy)” appear in search results?

You are doing a recording search, so you find recordings with that name. That is nothing to do with those aliases listed. You searched for something totally different with “(medley: Billy Jean / Jeopardy)”

The “alias” in your example is because someone uses “B. Project” and others “B-Project” and would not affect your search.

1 Like
  1. Primary locale aliases are used to support internationalization/localization. Not all languages are sorted in Latin script, or follow English conventions. Primary locale aliases should be entered all the time.
  2. Search hint aliases should be used if someone would reasonably search for what you’re using as the search hint. “Official” typos and common “romanizations” would make sense to add.

Trivia does not function as a search hint - e.g., it is not reasonable to attempt to lookup the label “EMI Classics” by typing out “Electric and Musical Industries Classics” (which has exactly 2 Google results, neither in English, if that gives you an idea of how likely it is somebody would try to do this and how unreasonable it would be in the context of music release metadata).

Aliases are created automatically after entity merges.

I’m of the opinion that it’s difficult to have too many aliases, both for localization (like on Super Smash Bros) and for searchability (like on Glaze/WoodenToaster or deadmau5). the more searchable an artist (or whatever) is, the better

the only exceptions are capitalization doesn’t matter (at least for Latin script, the one you’re likely reading right now), nor does punctuation, as MusicBrainz search treats most punctuation the same as a space (that said, it wouldn’t hurt to have these either). spaces (or lack thereof) do matter tho, which is why I’ve got both WoodenToaster and Wooden Toaster as aliases above

you may also notice I include common abbreviations too

4 Likes

It’s not difficult at all. False aliases pollute the search results for the entire database. You shouldn’t insert aliases with keywords that people wouldn’t reasonably use because it generates false positive matches that you have to page through before finding the thing you actually wanted. This is data quality 101 in all databases, not just MusicBrainz.

For example, “electric” currently contains several false matches that nobody should have a reasonable reason to ever use in the edit system or from normal searching. It’s caused by invalid aliases on those entries.

2 Likes

Thank you for saying that. Search results are already brittle enough on their own, we don’t do ourselves any favours by entering misleading aliases.

Another typical example: aliases that match another entry (legitimately entered) in the database, e.g. a legal name for an artist with multiple performance names, or a different performance name. It is always worth reminding other editors that Discogs aliases are a different species than MB’s ones.

1 Like

I said difficult, not impossible :sweat_smile:

(I didn’t even know EMI had a full legal name before this thread, just like CVS, lol)

that said, if an artist or whatever actually does/did use the full version of an acronym or initialism, we should probably add those, like TCB who went by TheClosetBrony when they started, but has since abbreviated, or maybe if a branch of EMI used to use their full name on releases

you’re talking like adding “Joel Zimmerman” or “TESTPILOT" to deadmau5? yeah, I can agree with you here

2 Likes

Yes, that would be reasonable. My example is intentionally one where it is not reasonable to have that alias because the labels at the top of the indexed search results have never gone by those names, and it shows how the false aliases bury actually relevant results far down the paging size of 10 at a time in the release editor.

1 Like

In case you weren’t aware, genres and tags broke indexed search.

1 Like