Intents with entities not matching

Does voiceflow’s prototype have a limited built-in NLP? When I use slot/entity extraction, it seems like only the slots at the very top of my list are trained, the rest require exact matches.



In testing, when I enter a fuzzy match for one of my first entities on the list, like campus life, it works.

“What is THE campus life and “WHAT’S campus life both match to the intent.

When I enter a fuzzy match for an entity way down on the list, like “study room” it doesn’t work. It only accepts EXACT matches to my utterances, like “What is (study room)

Any thoughts as to what’s happening?

Most NLUs need a prefix with the entity, using the entity alone is not recommended unless you really have to or if the interaction model is simple.

For this you can try using a Capture step instead as it’s configured to deal with {entity} alone.

You can also use the Test tool with the intent confidence log feature enabled, and you will be able to check what intent is catching the utterance and see if you have any conflicts.

Hi! I think there is a misunderstanding here. I don’t have any intents with only entities, I always have entity patterns inside examples like the ones above.
Example: “What is {topic}”

My issue is that the intents only capture the entity for the first 20 or so values in my list. Anything past that, the intent isn’t matching to the entity unless my input is an exact match for the sample utterance.

So for entity values 1-20(ish) I can say things like “what’s [financial aid]” or “what is the [writing center]” and it pulls correctly, even though they aren’t exact matches to my samples, but for entity values 21+ in my list for that entity, it only pulls the correct entity when it’s an exact match to my sample utterance.
Sample utterance: what is {topic}
What is [work study] —>exact match of sample, pulls “work study” correctly
What’s [work study] —> pulls no entity

This is one example, but my intents all have multiple sample utterances.