select * from page where page_namespace = 0 and page_is_redirect=0 and page_title regexp "^[A-Z0-9_\\W]*_\\([^a-z]*\\)$" 

and page_title not regexp "\\(([0-9]+_)?AM|FM|EP|TV|[0-9]+\\)$" -- Exclude various broadcasting stations per WP:NCBC

and page_title not regexp "\\([0-9]+([–-][0-9]+)?" -- Date based or hull number disambiguators are fine

and page_title not like "%C/20%" -- Exclude comet names per [[WP:NCASTRO]]. I disagree with this convention

and page_title not like "%C/19%" -- Exclude comet names per [[WP:NCASTRO]]. I disagree with this convention

and page_title not regexp "^VK_[0-9]+" -- Exclude some German tanks

and page_title not regexp "\\(ZVV|TILO\\)" -- Exclude some train lines that could probably have better names but whatever

and not exists (select 1 from pagelinks where pl_from=74115810 and pl_namespace=0 and pl_title=page_title) 

 -- Exclude things I've manually judged to be OK

order by page_title desc limit 1000;