I know we traditionally process tokens as case-insensitive, but... it seems reasonable to assume in HN comments that book titles would be capitalized properly (so we could ignore non-capitalized titles). Whether or not this information is present in the version on BigQuery, I'm not sure though.