-2

when i try to add utterance use name of entity "c_blankutterance"(Entity Type: Pattern.any), that failed about

"BadArgument: The entity extractor c_blankutterance does not exist in the application version. "

this error.

I checked that entity exist and train & publish is normaly working. and I added pattern used c_blankutterance before add other utterance. This reason is possible to failed?

BadArgument: The entity extractor c_blankutterance does not exist in the application version.

Nipun Tharuksha
  • 2,496
  • 4
  • 17
  • 40

1 Answers1

0

I suspect that you are misunderstanding how pattern.any is used. The pattern.any entity type is used specifically as a placeholder within a pattern utterance. Think of the placeholder as a kind of variable. As such, you can't assign this kind of entity type to a word/utterance in an intent.

If you are simply trying to match on specific words that might return as part of the "bad argument", then I would recommend you use a simple entity. If, however, you are unsure what might be returned, but the overall structure of the error will remain the same, then a pattern may serve you.

To use the pattern.any entity type, follow these steps:

First, (using your example utterance) create your intent (for example, "Arguments"). It doesn't need to contain any utterances, at this point.

Then, create the entity:

enter image description here

Next, navigate to the "Patterns" in the left menu, select the intent you created from the drop down list, and then type in your example utterance. Replace the word/words that may change with the entity you created, wrapped in curly braces ("{}"). This is the placeholder.

enter image description here

At this point, train your model and test. As you can see below, I entered in your example utterance substituting a random word where the placeholder sits. The result is a match on the entity.

enter image description here

Please note that using patterns will not guarantee correct intent mapping. Among other uses, patterns are meant to help improve prediction accuracy where scores are returning low.

You can read about patterns, as a concept, here and more on implementing here.

Hope of help.

Steven Kanberg
  • 6,078
  • 2
  • 16
  • 35