I'm looking out if there is a way where in we can track the missed utterances so that I can retrain the skill.
To be precise, we have LUIS(Microsoft's) and WATSON(IBM's) products available as part of cognitive services (I really apologize for getting other big guys in the market), we have a system where in, there are utterances linked to an Intent. for example there are 5 utterances for 2 intents, totaling to 10 utterances. And the user starts the conversation with the bot. Let's say that the user is talking to the bot with regards to the greeting Intent. and as part of the Intents-Utterances availability, in the background, we have
- Good morning
- Good Evening
- Good night
And these three are set to greeting Intent.
the user comes and says Good night. there is a default message given back(Since this is not part of the sample utterances given), In all three worlds(Alexa, Watson, LUIS).
Coming to the other cognitive services (MS and IBM), the admin(or any one who is authorized to enter the environment of that particular skill), just logs in and he can see the missed utterances there, he just needs to map this particular question(the missed one) to that corresponding intent, hit a retrain and publish. The next time the user says that question(Say Goodnight), this intent is caught and the appropriate response is delivered.
I'm wondering if we have a such similar approach in terms of Alexa. Can you please assist me on this.
This is one of the beautiful feature we are looking out for. So that the missed utterances of one user are no more missed utterances of another.
Please point me to the right direction.