According to the FAQ, I can sumbit corrections on artists and albums. What am I supposed to do if a song exists as a duplicate entry and the studio album links to one entry while a compilation album links to a different one?
Shall I sumbit a correction by choosing the entry that is featured on fewer albums and submit a correction on each album there, linking other entry in the comments?
This is a great question with a potentially frustrating answer.
We don't really have a terrific way to clean these different instances up.
The method you suggested is probably the best way, but I don't know if it would make it to the top of the data correction priority queue.
The songs in the database are not really grouped manually, they each exist as individual instances of tracks on each individual release.
We algorhythmically determine a "Song" or Performance ID by taking a look at the different metadata associated with each individual track.
This works well 99% of the time, but if one album lists a song by Joe Smith & Dave Thompson, but another album lists a song by J. Smith / D. Thompson (and yet another album lists the song by Smith &Thompson), different entries get created.
We have something like 35 million individual tracks in the database that boil down to 21 million individual performances of 16 million compositions.
There simply isn't the bandwidth to take a look at the original liner notes and how the record companies listed the composers to try to reconcile them to the same entries.
What is the specific instance you're looking at?
Hi Zac and thank you for your quick reply.
Your answer is far from frustrating: It was very explainative, fast (faster than mine) and I already imagined that the database would work like that. Manually grouping millions of songs is an impossible task.
I think a button to report duplicates, similar to the "Submit correction" button, would help a little there. Then a small notification at the top of the page pops up that links to a possible duplicate. People could vote on it and if it gets enough votes, the website will display a merged version for both instead, by just grabbing the metadata and reviews of both entries. By not merging the entries but just using links this can be easily reverted.
But I'm rambling again. That happens too often, sorry about that. I just get very passionate about ideas.
The song in question is Up on the Catwalk by Simple Minds. The two entries I have found are mt0051028515 and mt0008296390.
Thank you really much for this amazing website and community.
Yep. As expected, one instance has the proper songwriters and the other one is just listed as composed by Simple Minds.
We can think about some kind of community-based tool to indicate possible merges. Thanks for the suggestion.
