Closed tylerlittlefield closed 4 years ago
I think the tags are as expected. It is because the spacy model does not recognize "sam likes boat" as a sentence because everything is lowercased.
As seen in spacy_df()
part of speech tagging, ALL three words are tagged as proper nouns. So I guess spacy thinks "sam likes boat" is a name as a whole??
On the other hand, as long as "Sam" is properly capitalized, spacy parses the whole sentence correctly.
Either way, build_tree parses the data correctly based on spacy_df()
output and spacy_attributes()
values. Unless we want to prompt user as a reminder to reconsider casing in their input, I am not sure if there is a good way around this.
Got it, thanks for checking. I’ll close this issue!
For some reason, when the example "Sam likes boats" is lowercase, the root "likes" has attributes "PERSON" and "VERB". I think we would expect it to only have "VERB":
Also:
I checked out 28c79c145b0f96ddfe3e2e5d5ae2a7389cad34d2 with
git checkout 28c79c145b0f96ddfe3e2e5d5ae2a7389cad34d2
to double check if I broke things but the behavior seems to be the same.