Dataset Card for "relbert/conceptnet"

Dataset Summary

The selected subset of ConceptNet used in this work. 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.

• Number of instances

	train	validation	test
number of pairs	583082	1184	1187
number of unique relation types	28	20	19

• Number of pairs in each relation type

	number of pairs (train)	number of pairs (validation)	number of pairs (test)
AtLocation	69838	230	250
CapableOf	71840	124	144
Causes	34732	52	45
CausesDesire	9616	15	5
CreatedBy	534	1	2
DefinedAs	11048	2	1
DesireOf	28	0	0
Desires	8960	20	8
HasA	19234	43	41
HasFirstSubevent	7350	2	1
HasLastSubevent	5916	5	0
HasPainCharacter	2	0	0
HasPainIntensity	2	0	0
HasPrerequisite	47298	116	109
HasProperty	36610	63	70
HasSubevent	52468	82	83
InheritsFrom	112	0	0
InstanceOf	138	0	0
IsA	71034	197	211
LocatedNear	6	0	0
LocationOfAction	6	0	0
MadeOf	1518	10	14
MotivatedByGoal	23668	17	8
PartOf	5402	19	22
ReceivesAction	20656	15	11
RelatedTo	178	0	1
SymbolOf	328	2	0
UsedFor	84560	169	161

Dataset Structure

An example of train looks as follows.

{
  "relation": "IsA",
  "head": "baseball",
  "tail": "sport"
}

Citation Information

@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"
}