Datasets:

Languages:
English
Multilinguality:
monolingual
Size Categories:
10M<n<100M
Language Creators:
found
Annotations Creators:
found
Source Datasets:
original
ArXiv:
License:
article
string
link_text
string
is_same
uint8
A$AP Rocky
A$AP Rocky's
1
A$AP Rocky
ASAP Rocky
1
A$AP Rocky
Rakim Mayers
1
A$AP Rocky
Rakim Myers
1
A1000 road
A1000
1
A1000 road
High Road
1
A1013 road
A1013
1
A1018 road
A1018
1
A1020 road
A1020
1
A1023 road
A1023
1
A102 road
A102
1
A102 road
southern approach
1
A1033 road
A1033
1
A105 road
A105
1
A105 road
Wood Green High Road
1
A1065 road
A1065
1
A1065 road
A1065 Mildenhall to Fakenham road
1
A1067 road
A1067
1
A1067 road
A1067 Drayton Road
1
A1067 road
Drayton Road
1
A1068 road
A1068
1
A1077 road
A1077
1
A1079 road
A1079
1
A1079 road
Hull Road
1
A1081 road
A1081
1
A1081 road
A1081 dual carriageway
1
A10 autoroute
A10
1
A10 autoroute
A810
1
A10 autoroute
A837
1
A-10 Thunderbolt II
A-10
1
A-10 Thunderbolt II
A-10A
1
A-10 Thunderbolt II
A-10 and OA-10 Thunderbolt II
1
A-10 Thunderbolt II
A-10A Thunderbolt II
1
A-10 Thunderbolt II
A-10A Thunderbolt-II
1
A-10 Thunderbolt II
A-10A Thunderbolt IIs
1
A-10 Thunderbolt II
A-10A Thunderbolt II (Warthog)
1
A-10 Thunderbolt II
A-10C
1
A-10 Thunderbolt II
A-10C Thunderbolt II
1
A-10 Thunderbolt II
A-10C Thunderbolt II Ground Attack Aircraft
1
A-10 Thunderbolt II
A-10 ground attack jets
1
A-10 Thunderbolt II
A-10/OA-10
1
A-10 Thunderbolt II
A-10/OA-10 Thunderbolt II
1
A-10 Thunderbolt II
A-10 "Warthog"
1
A-10 Thunderbolt II
A-10s
1
A-10 Thunderbolt II
A-10 System Program Office
1
A-10 Thunderbolt II
A-10 Thunderbolt
1
A-10 Thunderbolt II
A-10 Thunderbolt II Warthog
1
A-10 Thunderbolt II
A-10 Thunderbolts
1
A-10 Thunderbolt II
A-10 Warthog
1
A-10 Thunderbolt II
A-10 Warthogs
1
A-10 Thunderbolt II
Air Force A-10
1
A-10 Thunderbolt II
A(later OA)-10 Thunderbolt II
1
A-10 Thunderbolt II
A/OA-10
1
A-10 Thunderbolt II
A/OA-10 Thunderbolt II
1
A-10 Thunderbolt II
A/OA-10 Thunderbolt IIs
1
A-10 Thunderbolt II
A-X ground attack aircraft
1
A-10 Thunderbolt II
Fairchild A-10
1
A-10 Thunderbolt II
Fairchild Republic A-10A Thunderbolt II
1
A-10 Thunderbolt II
Fairchild-Republic A-10/OA-10 Thunderbolt II
1
A-10 Thunderbolt II
Fairchild Republic A-10 Thunderbolt II
1
A-10 Thunderbolt II
Fairchild Republic<br /> A-10 Thunderbolt II Warthog
1
A-10 Thunderbolt II
Fairchild YA-10
1
A-10 Thunderbolt II
ground attack aircraft
1
A-10 Thunderbolt II
Republic A-10 Thunderbolt II
1
A-10 Thunderbolt II
Republic A/OA-10 Thunderbolt IIs
1
A1101 road
A1101
1
A1101 road
A1101 Bury St. Edmunds to Littleport road
1
A112 road
A112
1
A1134 road
A1134
1
A113 road
A113
1
A113 road
Chigwell Road
1
A1170 road
A1170
1
A1175 road
A1175
1
A1175 road
A175
1
A118 road
A118
1
A118 road
East London A118
1
A118 road
London Road, Romford
1
A118 road
main road
1
A118 road
Romford Road
1
A118 road
Stratford High Street
1
A1198 road
A1198
1
A11 autoroute
A 11
1
A11 autoroute
A11
1
A11 autoroute
A11 motorway
1
A11 autoroute
A844
1
A120 road
A120
1
A120 road
A120/
1
A120 road
A120 Braintree to Marks Tey
1
A120 road
A120 Hare Green to Harwich
1
A120 road
A120 trunk road
1
A1231 road
A1231
1
A1231 road
A1231)
1
A1237 road
A1237
1
A124 road
A124
1
A124 road
Barking Road
1
A124 road
St Mary's Lane
1
A124 road
Upminster Road
1
A127 road
A127
1
A127 road
A127 Southend Arterial Road
1
A127 road
A127 'Southend Arterial Road'
1

Wiki Entity Similarity

Usage:

from datasets import load_dataset

corpus = load_dataset('Exr0n/wiki-entity-similarity', '2018thresh20corpus', split='train')
assert corpus[0] == {'article': 'A1000 road', 'link_text': 'A1000', 'is_same': 1}

pairs = load_dataset('Exr0n/wiki-entity-similarity', '2018thresh20pairs', split='train')
assert corpus[0] == {'article': 'Rhinobatos', 'link_text': 'Ehinobatos beurleni', 'is_same': 1}
assert len(corpus) == 4_793_180

Corpus (name=*corpus)

The corpora in this are generated by aggregating the link text that refers to various articles in context. For instance, if wiki article A refers to article B as C, then C is added to the list of aliases for article B, and the pair (B, C) is included in the dataset.

Following (DPR https://arxiv.org/pdf/2004.04906.pdf), we use the English Wikipedia dump from Dec. 20, 2018 as the source documents for link collection.

The dataset includes three quality levels, distinguished by the minimum number of inbound links required to include an article in the dataset. This filtering is motivated by the heuristic "better articles have more citations."

Min. Inbound Links Number of Articles Number of Distinct Links
5 1,080,073 5,787,081
10 605,775 4,407,409
20 324,949 3,195,545

Training Pairs (name=*pairs)

This dataset also includes training pair datasets (with both positive and negative examples) intended for training classifiers. The train/dev/test split is 75/15/10 % of each corpus.

Training Data Generation

The training pairs in this dataset are generated by taking each example from the corpus as a positive example, and creating a new negative example from the article title of the positive example and a random link text from a different article.

The articles featured in each split are disjoint from the other splits, and each split has the same number of positive (semantically the same) and negative (semantically different) examples.

For more details on the dataset motivation, see the paper. If you use this dataset in your work, please cite it using the ArXiv reference.

Generation scripts can be found in the GitHub repo.

Downloads last month
7
Edit dataset card