idafensp / nlp4types Goto Github PK

Hi, I found a small bug in the tool when deleting rows containing owl:Thing as type.

In the function get_types_and_abstracts of the main.py file there is a call to the function get_resources_from_types with 3 arguments: class list, path of the types file and boolean encode types ([], types_path, enctyp) but the function y accepts 4 arguments: class list, path of the types file, boolean remove owl:Thing and boolean encode types ([], types_path, enctyp).

NLP4Types/main.py

Line 86 in 13cc8a7

dft = dbp.get_resources_from_types([], types_path, enctyp)

NLP4Types/utils/dbpedia.py

Line 31 in 13cc8a7

def get_resources_from_types(class_list, db_types_file, remove_owl_thing = True, encode=False):

The problem is that the last argument of the call, enctyp, goes to the third argument of the function that corresponds to remove_owl_thing.
I noticed this when I saw that there were instances of owl:thing in both predicates and labels (ground truth).

A possible solution would be to change dft = dbp.get_resources_from_types([], types_path, enctyp) to dft = dbp.get_resources_from_types([], types_path,remove_owl_thing=True, encode=enctyp) in get_resources_from_types()

It seems that this bug can affect the result, making it worse.