For now when entering a classical release you have to encode every track name using a very complicated ClassicalStyleGuide . If you have another interpretation of the same work you have to do everything again.
Every track name contains a mix of different data that a computer can parse only if the data is correctly encoded (which is rare). A more sober approach is to create a database with the list of works of every classical composer (ClassicalBrainz) and link each track from MusicBrainz with the correct entry in ClassicalBrainz.
Some people already described a possible implementation.