Part of Speech as conditional patterns

Hi Team,

I have been looking for this concrete functionality but Im not sure if it is possible to achive. While analizing the bot behaviour through the NL Analysis window provided, it does give info related to the Part of Speach, kind of:

“pos”: "Noun_infinitive Verb_unknown "

I would like to know if theres a way to use this pos components as conditional patterns for utterances, entities, etc… And if possible would it be aplied also to synonims?

Thanks for your time,
Roberto

Great question, :grin:
Because yes there is a way!

Under the covers, the net result of the POS tagging is to assign (or “marking” as it is called) concepts to every word (or phrase) in the utterance. Each POS tag has its own internal, dynamic, concept, and actually there is a hierarchy of them.

So a word that is tagged as a verb will be marked as ~verb and also possible one of ~verb_present, ~verb_present_participle, ~verb_past, ~verb_infinitive, or a few others. Verb_unknown actually means that it cannot determine any tense to mark other than plain ~verb.

Nouns are ~noun, but there are also ~noun_infinitive (for when a verb is used as a noun), ~noun_gerund, ~noun_singular, etc.

And as you should know, concepts can be used in patterns - however note that these are dynamic concepts in that there are no predefined contents and so they cannot be used in the areas which require training (e.g. trait patterns).

I’m going to work with the technical writers to see if we can get a list of these published.

1 Like

Awesome!

Thanks for the help and the quick reply!