Utterances as they stand now have no way of expressing which slots contain and where they are, some way of identifying this is needed.
Some possibilities:
As special characters inside the utterance itself: "Hello and welcome to ${place}". This needs escaping if we want them to be shown.
Having the utterance broke down into pieces: [ {text: "Hello and welcome to "}, {slot: "place"} ]. In this form we can add more data into
Having the slots indicated as positions: { text: "Hello and welcome to my place", slots: [{name:"place", start:21, end: 29}]. This form might be faster when translating into some NLUs (as they need an example) but it is pretty cumbersome to implement.
Utterances as they stand now have no way of expressing which slots contain and where they are, some way of identifying this is needed.
Some possibilities:
"Hello and welcome to ${place}"
. This needs escaping if we want them to be shown.[ {text: "Hello and welcome to "}, {slot: "place"} ]
. In this form we can add more data into{ text: "Hello and welcome to my place", slots: [{name:"place", start:21, end: 29}]
. This form might be faster when translating into some NLUs (as they need an example) but it is pretty cumbersome to implement.