Lets say I wanted to autogenerate questions for a wikipedia page as a sort of comprehension exercise, how would i best go about that.

Using machine learning would be necessary but any other things i need to look into?

  • For fill in the blank questions scan for x is y statements. Turn x or y into a blank. For multiple choice or true false either quote directly or corrupt. This part is easy. The hard part is discovering which ones are obvious and which are deceptive. – candied_orange Sep 27 '20 at 14:31

I did some work on code from Michael Heilmann and Noah H. Smith which attempts to tackle this issue (question generation from text), and can be used as starting point for further research:


This is from around 2010, so 10 years old. There has probably been more research since then, maybe this can help to get started.

What I did to actually generate questions was very basic, and is not the right way to do it (IIRC I replaced the part of the sentence I wanted to ask about with a question word, then permutated the sentence until I got one that successfully parsed as a question), so only use this as demonstration. It also uses Wikipedia as a source so that is also useful to see, if you are using Java:


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.