EDIT

For nomenclatural references we do not necessarily need to store the author together with the "article". It is clear from the context that the author of the name is the author in the volume.
We could have records representing a volume of a journal. This reduces the number of records a lot as we do not need to distinguish records for each author. As the author is usually the only information that differs between articles.
If one wants to link to a concrete article or a Section in an articles this is still possible. Only the formatter needs to support it, but already does.

Also we need to discuss if we want to have 2 different types, volumes and articles and if yes, if articles should always have volumes as inRef. In later case we have up to 3 inRefs: section, article, volume, journal for clearly define a section of an article.

For book sections it is more difficult because book sections inherit more information from their book so it does not make much sense to give them an own record which differs NOT in authorship. For nom. references it might be possible in most cases to directly link to the book and only decide via name authorship comparison if it is a book section or the book itself. Question: are book section possible having the exact same author as the book? Another problem is 2 different records representing the same author may lead to incorrect comparison, especially as book authors might not be abbreviated. So here we probably should keep the current system.