rloganiv / kglm-data

Code used to create the Linked WikiText-2 dataset
17 stars 2 forks source link

what does 'KG' mean in source field? #3

Closed jcyk closed 4 years ago

jcyk commented 4 years ago

hi, thank you for open-sourcing this code.

I have a question regarding the annotation. I see WIKI NEL COREF and KG for annotation source. what does KG stand for in the annotations?

rloganiv commented 4 years ago

Hi @jcyk,

KG stands for knowledge graph. It indicates the link was created from an exact string match to an entity's alias in the local knowledge graph.

jcyk commented 4 years ago

hi @rloganiv thanks for your quick reply.

could you please explain more about the difference between WIKI and KG? I am kind of confused.

Thanks!

rloganiv commented 4 years ago

WIKI means the link comes from directly from the article HTML. KG means the link comes from a knowledge graph (in our case WikiData).

rloganiv commented 4 years ago

I've added a description to all of the fields to the README. More details about how the KG links are produced are included in our paper. The relevant pieces of code are in the annotate.py file.