HyperAI초신경

홈 뉴스 최신 연구 논문 튜토리얼 데이터셋 백과사전 SOTA LLM 모델 GPU 랭킹 컨퍼런스

한국어

HyperAI초신경

Aspect Based Sentiment Analysis Absa On Aste

평가 지표

F1 (L14)

F1 (R15)

F1 (R16)

F1(R14)

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름	F1 (L14)	F1 (R15)	F1 (R16)	F1(R14)	Paper Title	Repository
MvP (multi-task)	65.30	69.44	73.10	76.30	MvP: Multi-view Prompting Improves Aspect Sentiment Tuple Prediction
Paraphrase	61.13	62.56	71.70	72.03	Aspect Sentiment Quad Prediction as Paraphrase Generation
ChatGPT (gpt-3.5-turbo, few-shot)	38.12	-	-	-	MvP: Multi-view Prompting Improves Aspect Sentiment Tuple Prediction
AugABSA	62.66	65.80	74.23	73.76	Generative Data Augmentation for Aspect Sentiment Quad Prediction
GAS	58.19	60.23	69.05	70.52	Towards Generative Aspect-Based Sentiment Analysis
ChatGPT (gpt-3.5-turbo, zero-shot)	36.05	-	-	-	MvP: Multi-view Prompting Improves Aspect Sentiment Tuple Prediction
DLO	61.46	64.26	73.03	72.39	Improving Aspect Sentiment Quad Prediction via Template-Order Data Augmentation
ASTE-Transformer	64.90	72.10	76.19	76.06	ASTE Transformer Modelling Dependencies in Aspect-Sentiment Triplet Extraction
MvP	63.33	65.89	73.48	74.05	MvP: Multi-view Prompting Improves Aspect Sentiment Tuple Prediction
Seq2Path	64.82	65.88	72.87	75.52	Seq2Path: Generating Sentiment Tuples as Paths of a Tree	-
LEGO-ABSA (multi-task)	62.20	64.40	69.90	73.70	LEGO-ABSA: A Prompt-based Task Assemblable Unified Generative Framework for Multi-task Aspect-based Sentiment Analysis	-
UIE	62.94	64.41	72.86	72.55	Unified Structure Generation for Universal Information Extraction
Span-ASTE	59.38	63.27	70.26	71.85	Learning Span-Level Interactions for Aspect Sentiment Triplet Extraction

0 of 13 row(s) selected.