Hi,
suppose we have a movie database and a rating system based on 3 grades “good”, “ok”, “bad”. What’s the best approach to model the intents/entities in a chatbot? I thought of the following cases:
Case 1: one intent, one entity rating: the user might for example say it’s a good movie! . Then good will be recognised as a rating. But what if the user says I liked the movie? I thought we might treat “liked” as synonym for “good” but it quickly gets messy since I don’t want to define every possible synonym word or similar phrase to “good” as a synonym, especially if I have other entities where a given word (e.g. fantastic) should not be mapped to good.
Case 2: three intents, no entities. One intent for each grade, handles perfectly utterances such as “I liked the movie!”, no need to define synonyms. Drawback: number of intents grows linearly with the number of possibilities, instead of having one intent / one entities.
Any suggestions or ideas how to model this?
Thanks, Amine