relbert/relbert-roberta-base-nce-conceptnet
Feature Extraction
•
Updated
•
231
•
1
The viewer is disabled because this dataset repo requires arbitrary Python code execution. Please consider
removing the
loading script
and relying on
automated data support
(you can use
convert_to_parquet
from the datasets
library). If this is not possible, please
open a discussion
for direct help.
The selected subset of ConceptNet used in this work, which compiled
to fine-tune RelBERT model.
We removed NotCapableOf
and NotDesires
to keep the positive relation only.
We consider the original test set as test set, dev1 as the training set, and dev2 as the validation set.
An example of train
looks as follows.
{
"relation_type": "AtLocation",
"positives": [["fish", "water"], ["cloud", "sky"], ["child", "school"], ... ],
"negatives": [["pen", "write"], ["sex", "fun"], ["soccer", "sport"], ["fish", "school"], ... ]
}
train | validation | test |
---|---|---|
28 | 34 | 16 |
@InProceedings{P16-1137,
author = "Li, Xiang
and Taheri, Aynaz
and Tu, Lifu
and Gimpel, Kevin",
title = "Commonsense Knowledge Base Completion",
booktitle = "Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) ",
year = "2016",
publisher = "Association for Computational Linguistics",
pages = "1445--1455",
location = "Berlin, Germany",
doi = "10.18653/v1/P16-1137",
url = "http://aclweb.org/anthology/P16-1137"
}