3

This should be a simple question to many in the area, but for the life of me I can't find a solution by myself.

I need a corpus of medical text for the purposes of training a spell checker which will be used on... medical text. For this I need a large amount of text that has few (or ideally no) errors in it. The abstracts from Medline articles would be a good bet.

However, while I can many solutions that are grossly more complicated than I require (such as treebank) designed for tree-building or POS tasks, and would require significant work to remove annotation and compile into a single .txt, surely such a resource already exists?

Alternatively I could just visit a hundred or so Medline articles and start copy-pasting their abstract details I suppose...

migrated from datascience.stackexchange.com Feb 21 '16 at 15:45

This question came from our site for Data science professionals, Machine Learning specialists, and those interested in learning more about the field.

  • What do you want to spell check? medical research articles, or physician notes? – Franck Dernoncourt Feb 21 '16 at 16:10
  • @FranckDernoncourt it would be physician notes: so terse, and domain specific term heavy. Lots of acronyms as well, but I'm handling those separately (for the purposes of spell-checking, simply ignoring them) – Stumbler Feb 21 '16 at 17:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.