Skip to content

Part-of-speech tagging

Standard tagging

Standard part-of-speech tagging is the deep linguistic analysis process that marks up each token with the corresponding Universal POS tag.

For example, for this sentence:

Michael Jordan was one of the best basketball players of all time.

standard part-of-speech tagging produces this output:

Token Part-of-speech Universal POS tag
Michael Jordan Proper noun PROPN
was Auxiliary AUX
one Numeral NUM
of Adposition ADP
the Determiner DET
best Adjective ADJ
basketball players Noun NOUN
of Adposition ADP
all Determiner DET
time Noun NOUN
. Punctuation PUNCT

Standard part-of-speech tagging output is part of the JSON object returned by deep linguistic analysis.

Custom tagging

In addition to standard part-of-speech tagging, deep linguistic analysis marks up both tokens and atoms with custom expert.ai type labels.
expert.ai types combine part-of-speech information with entity type information.
For example, for the following sentence:

Please Travis, take me to Avalon. Do you mind if I bring my dog Bella with me?

custom tagging is:

Token Type description Custom expert.ai label
Please Adverb ADV
Travis Proper name of a human being NPR.NPH
, Punctuation PNT
take Verb VER
me Pronoun PRO
to Preposition PRE
Avalon Proper noun of an extra-terrestrial or imaginary place GEX
. Punctuation PNT
Do Auxiliary verb AUX
you Pronoun PRO
mind Verb VER
if Conjunction CON
I Pronoun PRO
bring Verb VER
my Adjective ADJ
dog Noun NOU
Bella Proper noun of an animal NPR.ANM
with Preposition PRE
me Pronoun PRO
? Punctuation PNT

As mentioned above, the expert.ai type is also attributed to atoms, while standard part-of-speech tagging stops at the token level.

Custom part-of-speech tagging output is part of the JSON object returned by deep linguistic analysis.