PK PyTorchKR Papers
Papers with Code · CC BY-SA 4.0

Relation Extraction

552 평가 기록 101 datasets

Relation Extraction is the task of predicting attributes and relations for entities in a sentence. For example, given a sentence “Barack Obama was born in Honolulu, Hawaii.”, a relation classifier aims at predicting the relation of “bornInCity”. Relation Extraction is the key component for building relation knowledge graphs, and it is of crucial significance to natural language processing applications such as structured search, sentiment analysis, question answering, and summarization.

<span class="description-source">Source: Deep Residual Learning for Weakly-Supervised Relation Extraction </span>

개요는 각 항목의 일부입니다. 위 탭에서 전체 기록을 페이지별로 탐색할 수 있습니다.

벤치마크

Papers with Code 아카이브의 데이터셋별 기록 예시입니다. 평가 조건과 지표 방향이 검증되지 않아 순위를 의미하지 않습니다.

Dataset기록된 모델 예시MetricPaperCode
DREEAM
F1: 67.53Ign F1: 65.47
논문 코드
RAG4RE
F1: 86.6
논문 코드
ITER
RE Micro F1: 75.1 ± 0.49NER Micro F1: 91.6 ± 0.12RE+ Micro F1: 71.9 ± 0.56Sentence Encoder: FLAN T5 3B
논문 코드
SP
F1: 91.9
논문 -
GPT-4o (CoT)
Text Score: 59.2Video Score: 51Group Score: 35
- -
HiDialog
F1c (v2): 68.2F1 (v2): 77.1
논문 코드
Deepstruct zero-shot
F1: 36.1
논문 코드
REBEL
RE+ Micro F1: 75.4RE+ Macro F1 : 76.65
논문 코드
ITER
RE+ Macro F1: 85.6 ± 1.42NER Macro F1: 92.63 ± 0.89
논문 코드
UniRel
F1: 94.7
논문 코드
SciBert (Finetune)
F1: 83.64
논문 코드
PL-Marker
RE Micro F1: 69.7NER Micro F1: 90.4RE+ Micro F1: 66.5Cross Sentence: Yes
논문 코드

데이터셋

101개
DocREDTACREDACE 2005SemEval-2010 Task-8VinogroundDialogRETACREDCoNLL04Adverse Drug Events (ADE) CorpusWebNLGChemProtACE 2004

논문

라이브러리

Papers with Code 리더보드 row에 연결된 구현 저장소입니다.

github.com/pkuserc/rela 1 논문
github.com/amazon-science/e2e-docie 1 논문
github.com/klimzaporojets/consistent-el 1 논문
github.com/klimzaporojets/DWIE 1 논문
github.com/klimzaporojets/e2e-kb-ie 1 논문
github.com/slczgwh/REDN 1 논문
github.com/UKPLab/emnlp2017-relation-extraction 1 논문
github.com/zhao9797/RIFRE 1 논문

하위 태스크

JE Joint Entity and Relation Extraction 16 데이터셋 · 54 rows RC Relation Classification 12 데이터셋 · 37 rows TR Temporal Relation Extraction 5 데이터셋 · 37 rows DR Dialog Relation Extraction 2 데이터셋 · 18 rows RE Relationship Extraction (Distant Supervised) 2 데이터셋 · 10 rows DR Document-level Relation Extraction 6 데이터셋 · 9 rows ZR Zero-shot Relation Triplet Extraction 2 데이터셋 · 5 rows HE Hyper-Relational Extraction 1 데이터셋 · 4 rows BR Binary Relation Extraction 2 데이터셋 · 3 rows 4R 4-ary Relation Extraction 1 데이터셋 · 1 rows D DrugProt 1 데이터셋 · 1 rows CR Continual Relation Extraction 0 데이터셋 · 0 rows