On Saturday, March 14, 2020 at 9:39:15 AM UTC-4, Thomas Passin wrote: > > On Saturday, March 14, 2020 at 7:33:28 AM UTC-4, Edward K. Ream wrote: >> >> >> Otoh, Leo's Qt dock code depends on complex state set on program startup >> and shutdown, as well as arbitrary *sequences* of user actions. I agree >> that the code is hard to test and hard to use. That does not mean that >> testing that code will ever be easy. If anyone knows how to test the Qt >> code, I would appreciate knowing how. >> > > On my main programming gig, I develop and support an application, similar > to but not the same as assigning medical diagnostic codes. Given a query, > the system has to produce X number of candidate codes and score them. > Trouble is, there is no purely objective way to say definitively which > responses are "correct". Some are obviously wrong, but it's for the most > part a human judgement call. > > Any minor change to the ranking algorithm produces changes in the > results. But I can't automate testing because there is no correct or even > standard answer. The best that I've figured out is to have a person rate > the results based on several subjective criteria. And to be thorough, you > need to rate at least 300 query results to get good enough statistics to > show an improvement. >
-- You received this message because you are subscribed to the Google Groups "leo-editor" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/leo-editor/b18c0658-7b56-42f6-9d0f-a36c619df048%40googlegroups.com.
