Music Release Disambiguation using Multi-Modal Neural Networks

A system comprising:

• one or more processors;
• and
• memory storing instructions that, when executed by the one or more processors, cause the system to carry out operations comprising:
• providing, to an artificial neural network (ANN) implemented on a computing system, musical-artist-release records (MARRs) each being associated with a musical release of an artist and comprising (i) a release ID of the musical release and a unique artist identifier (ID) of the artist and (ii) release data assigned according to release-data categories that comprise particular music media content associated with the musical release, and a set of metadata categories including a sub-definitive musician name of the artist and a subset of release subcategories specific to the musical release;
• forming distinct groupings of the MARRs;
• for each grouping of the distinct groupings of the MARRs, applying the ANN to each MARR of the grouping to generate a release feature vector (RFV) for each MARR of the grouping wherein the RFV is associated with the unique artist ID of the MARR and comprises a set of sub-feature vectors that characterize the release data of the release-data categories of the MARR, and wherein the RFV is defined within an RFV space;
• for each grouping of the distinct groupings of the MARRs, training the ANN and adjusting the RFV to cluster in the RFV space with other RFVs associated with the same unique artist ID, and to separate in the RFV space RFVs associated with different unique artist IDs;
• and
• storing each of the MARRs together with its respective RFV in a musical release database.

A non-transitory computer-readable medium having instructions stored thereon that, when executed by one or more processors of a system for disambiguating musical artist names, cause the system to carry out operations including:

• providing, to an artificial neural network (ANN) implemented on a computing system, musical-artist-release records (MARRs) each being associated with a musical release of an artist and comprising (i) a release ID of the musical release and a unique artist identifier (ID) of the artist and (ii) release data assigned according to release-data categories that comprise particular music media content associated with the musical release, and a set of metadata categories including a sub-definitive musician name of the artist and a subset of release subcategories specific to the musical release;
• forming distinct groupings of the MARRs;
• for each grouping of the distinct groupings of the MARRs, applying the ANN to each MARR of the grouping to generate a release feature vector (RFV) for each MARR of the grouping wherein the RFV is associated with the unique artist ID of the MARR and comprises a set of sub-feature vectors that characterize the release data of the release-data categories of the MARR, and wherein the RFV is defined within an RFV space;
• for each grouping of the distinct groupings of the MARRs, training the ANN and adjusting the RFV to cluster in the RFV space with other RFVs associated with the same unique artist ID, and to separate in the RFV space RFVs associated with different unique artist IDs;
• and
• storing each of the MARRs together with its respective RFV in a musical release database.