It would be a better idea to self populate the hashes with the information that could have been scraped from wherever the ISBN data is obtained (i.e. Amazon) as it already has a list of categories. Not sure that auto-populating for the same author only makes as much general sense.

Good idea!