This repository is used to compute a 5-star rating for the MusicBrainz song dataset. You can read the formula in the code.
Note
You can use this script along with the new Edit Documents with Function prototype to do it more easily.
- First compile the program in release.
cargo build --release- Once you downloaded the NDJSON MusicBrainz dataset you can pass it a first time in this program to output documents with a raw-rated field.
gzcat ~/Downloads/musicbrainz.ndjson.gz | ./target/release/musicbrainz-artist-scoring | gzip | cpipe -vt > ~/Downloads/musicbrainz-rated.ndjson.gz- After this you'll need to run a small script that extracts the raw-rated scores for the top musics of the dataset. This script will output the
BEST_SCOREandSMALLEST_TOP_SCOREthat you'll need to modify in the code.
./compute-best-scores.sh- Once you modified then, please go to step 1 and stop before step 3. You can use the latest dataset you just got with a great 5-star rating included 🎉