notesum.ai

Published at October 18

Few-Shot Joint Multimodal Entity-Relation Extraction via Knowledge-Enhanced Cross-modal Prompt Model

cs.CL

cs.AI

cs.LG

Released Date: October 18, 2024

Authors: Li Yuan¹, Yi Cai¹, Junsheng Huang¹

Aff.: ¹South China University of Technology, Guangzhou, China

Arxiv: https://arxiv.org/abs/2410.14225v1

Refer to caption

Modality	Model	Micro			Macro
Modality	Model	P	R	F₁	P	R	F₁
Text	Tplinker	40.03 $\pm$ (2.51)	22.91 $\pm$ (3.67)	29.00 $\pm$ (2.86)	19.40 $\pm$ (2.35)	8.94 $\pm$ (1.56)	12.21 $\pm$ (2.04)
	BiRTE $\clubsuit$	50.69 $\pm$ (3.19)	26.95 $\pm$ (3.52)	35.10 $\pm$ (3.43)	23.89 $\pm$ (4.44)	11.74 $\pm$ (1.48)	15.70 $\pm$ (1.78)
	SpERT $\clubsuit$	44.27 $\pm$ (2.13)	32.78 $\pm$ (4.51)	37.69 $\pm$ (3.02)	25.19 $\pm$ (2.35)	16.05 $\pm$ (2.03)	19.60 $\pm$ (1.78)
Text-Image	AGBAN+MEGA	41.45 $\pm$ (2.87)	27.02 $\pm$ (3.17)	32.71 $\pm$ (3.02)	24.71 $\pm$ (4.17)	10.87 $\pm$ (1.34)	15.10 $\pm$ (2.02)
	UMGF+MEGA	42.31 $\pm$ (3.25)	25.13 $\pm$ (1.97)	31.53 $\pm$ (2.73)	23.09 $\pm$ (4.01)	9.98 $\pm$ (2.35)	13.94 $\pm$ (3.57)
	EEGA $\spadesuit$	47.27 $\pm$ (3.28)	29.65 $\pm$ (2.65)	36.44 $\pm$ (2.89)	25.23 $\pm$ (3.35)	13.31 $\pm$ (2.56)	17.43 $\pm$ (2.89)
	KECPM	44.06 $\pm$ (1.46)	36.55 $\pm$ (1.88)	39.96 $\pm$ (1.73)	24.45 $\pm$ (2.12)	18.85 $\pm$ (2.49)	21.29 $\pm$ (2.34)