Learning High-Quality and General-Purpose Phrase Representations
Paper
• 2401.10407 • Published
Dataset stringlengths 4 28 | Left int64 201 6.93k | Right int64 10 1.16k | Matches int64 10 1.16k |
|---|---|---|---|
Amphibian | 3,663 | 1,161 | 1,161 |
ArtificialSatellite | 1,801 | 72 | 72 |
Artwork | 3,112 | 245 | 245 |
Award | 3,380 | 384 | 384 |
BasketballTeam | 928 | 166 | 166 |
Case | 2,474 | 380 | 380 |
ChristianBishop | 5,363 | 494 | 494 |
ClericalAdministrativeRegion | 2,547 | 190 | 190 |
Country | 2,791 | 291 | 291 |
Device | 6,933 | 658 | 658 |
Drug | 5,356 | 157 | 157 |
Election | 6,565 | 727 | 727 |
Enzyme | 3,917 | 48 | 48 |
EthnicGroup | 4,317 | 946 | 946 |
FootballLeagueSeason | 4,457 | 280 | 280 |
FootballMatch | 1,999 | 53 | 53 |
Galaxy | 555 | 17 | 17 |
GivenName | 3,021 | 154 | 154 |
GovernmentAgency | 3,977 | 571 | 571 |
HistoricBuilding | 5,064 | 512 | 512 |
Hospital | 2,424 | 257 | 257 |
Legislature | 1,314 | 216 | 216 |
Magazine | 4,005 | 274 | 274 |
MemberOfParliament | 5,774 | 503 | 503 |
Monarch | 2,033 | 242 | 242 |
MotorsportSeason | 1,465 | 388 | 388 |
Museum | 3,982 | 305 | 305 |
NCAATeamSeason | 5,619 | 34 | 34 |
NationalFootballLeagueSeason | 3,003 | 10 | 10 |
NaturalEvent | 970 | 51 | 51 |
Noble | 3,609 | 364 | 364 |
PoliticalParty | 5,254 | 495 | 495 |
Race | 2,382 | 175 | 175 |
RailwayLine | 2,189 | 298 | 298 |
Reptile | 666 | 819 | 562 |
RugbyLeague | 418 | 58 | 58 |
ShoppingMall | 201 | 227 | 159 |
SoccerClubSeason | 1,197 | 51 | 51 |
SoccerLeague | 1,315 | 238 | 238 |
SoccerTournament | 2,714 | 290 | 290 |
Song | 5,726 | 440 | 440 |
SportFacility | 6,392 | 672 | 672 |
SportsLeague | 3,106 | 481 | 481 |
Stadium | 5,105 | 619 | 619 |
TelevisionStation | 6,752 | 1,152 | 1,152 |
TennisTournament | 324 | 27 | 27 |
Tournament | 4,858 | 459 | 459 |
UnitOfWork | 2,483 | 380 | 380 |
Venue | 4,079 | 384 | 384 |
Wrestler | 3,150 | 464 | 464 |
Learning High-Quality and General-Purpose Phrase Representations.
Lihu Chen, Gaël Varoquaux, Fabian M. Suchanek.
Accepted by EACL Findings 2024
Our PEARL Benchmark contains 9 phrase-level datasets of five types of tasks, which cover both the field of data science and natural language processing.
| - | PPDB | PPDB filtered | Turney | BIRD | YAGO | UMLS | CoNLL | BC5CDR | AutoFJ |
|---|---|---|---|---|---|---|---|---|---|
| Task | Paraphrase Classification | Paraphrase Classification | Phrase Similarity | Phrase Similarity | Entity Retrieval | Entity Retrieval | Entity Clustering | Entity Clustering | Fuzzy Join |
| Samples | 23.4k | 15.5k | 2.2k | 3.4k | 10k | 10k | 5.0k | 9.7k | 50 subsets |
| Averaged Length | 2.5 | 2.0 | 1.2 | 1.7 | 3.3 | 4.1 | 1.5 | 1.4 | 3.8 |
| Metric | Acc | Acc | Acc | Pearson | Top-1 Acc | Top-1 Acc | NMI | NMI | Acc |
from datasets import load_dataset
turney_dataset = load_dataset("Lihuchen/pearl_benchmark", "turney", split="test")
We offer a python script to evaluate your model: eval.py
python eval.py -batch_size 32
@article{chen2024learning,
title={Learning High-Quality and General-Purpose Phrase Representations},
author={Chen, Lihu and Varoquaux, Ga{\"e}l and Suchanek, Fabian M},
journal={arXiv preprint arXiv:2401.10407},
year={2024}
}