i want to train luis ai with sufficient utterances uploaded through luis api -
i new luis ai.
train luis bot users wants buy books online. possible enter i want xyz
, xyz book or i want abc
, abc author. can write find
, find out
, search
, searching
, looking
, would see
, would find
or want write.
my requirement begin excel-sheet utterances , entities , when upload it, click on train, application should trained enough handle such user input, @ least 90%.
the problem here how should write utterances handle huge probability of user input. have approx 65 utterances includes relevant , diverse utterance still not getting trained handle user input.
please suggest me how proceed utterances meet requirement.
scientist take 30 minutes of conversation or 200 utterances enough sample conduct research [1] order of magnitude estimate know , compare ourselves to.
now, variability of incoming utterances, 1 must find origin of similar requests. case, sites yahoo answers great finding usual structure of requests on topic work into. suggest find place people query similar objective: google adwords helper general solid start.
[1] http://www.scielo.br/scielo.php?pid=s1516-18462015000401143&script=sci_arttext&tlng=en
Comments
Post a Comment