If we autolearn, need to gather as many identifiers as possible to see if we match something already in the database.
Undoubtedly this should go at a slightly lower level in terms of normalization, massaging, dup-checking, etc. before we insert new pubs into the database.
If we autolearn, need to gather as many identifiers as possible to see if we match something already in the database.
Undoubtedly this should go at a slightly lower level in terms of normalization, massaging, dup-checking, etc. before we insert new pubs into the database.