Open elliotclee opened 2 months ago
Hello @elliotclee,
This another great idea. I know when I do an exercise on Duo Lingo, the spelling errors are ignored, we should have the same functionality. Thank you also for linking those resources.
Sincerely, Timothy
Currently the vocabulary auto-grading module uses a pretty inflexible method of checking students' answers. In order to accommodate even basic variations like British vs USA spellings, the variations have to individually be added to the vocab database, which still has errors. As a result, it is easy for students to have questions marked wrong even though they fundamentally know the correct answer. It would benefit students if some work could be done to improve this situation.
There are phonetic matching algorithms out there that could be used on students' answers to accommodate minor spelling mistakes, spelling variations, etc. My understanding is that they are language-specific, so they are not a universal solution, but if a good framework is put in place now, it will be easy to add other matching algorithms such as AI-based ones in the future.
Three major phonetic matching algorithms that I'm aware of are Soundex, NYSIIS, and Metaphone. See https://digitalcommons.uri.edu/srhonorsprog/882/ for some helpful info.