With regards to the Filter Search Api Endpoint:
Is the API able to handle simple typos and misspellings? I’m certain the answer is ‘yes’ however, I’d like to better understand the limits of what the API is capable of in terms of
Spell Check
Typo Tolerance
Semantic Understanding
etc.
Thanks for your help!
Hey Lenny -
The filter search endpoint has typo tolerance built-in, but it doesn’t do spell checking or semantic understanding. The difference between typo tolerance and spell checking is that typo tolerance simply returns all words within a certain edit distance of the query, whereas spell checking uses a statistical model to predict whether a word was misspelled and, if so, what the intended word was.
Additionally, filter search doesn’t do any sort of semantic understanding the way the main search algorithm does. The reason for these limitations has mostly to do with speed - running semantic NLP models adds extra latency, but this endpoint is designed to run with each keystroke, so it needs to be lightning fast.
Hope this helps!
Max
