Datasets:

patriziobellan

/

PET

like 14

First of all: Thank you for making this dataset freely available, it's an excellent source for my own work.

When processing the dataset, I encountered a possibly problematic relation in "doc-2.1". Here you have tagged the tokens "problem", "fix", "process" (sentence 6, tokens 14-16) as "B-Activity Data", "I-Activity Data", "O" respectively. A relation points to token 16 in sentence 6 ("process"), i.e. relation "uses" with source (sent. 3, token 11) and target (sent. 6, token 16).

This is the only occurence, where a "O" token is target/source of a relation. Is there a reason, why you annotated it this way?

I would have expected the span "problem", "fix", "process" to be labeled "B-Activity Data", "I-Activity Data", "I-Activity Data" and the relation have the target (sent. 6, token 14), i.e. token "fix".

Looking forward to hearing from you!