id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
fcis-30522	Deng, Linlin; Lu, Xina	Study on Small Sample Text Classification Based on Multi-Level Self-Attention and Multi-Feature Residual Fusion under Data Enhancement	2025	9	.pdf	application/pdf	4426	280	53	Additionally, the residual feature fusion methodology preserves critical original information while markedly improving model performance. As shown in (Fig 4), comparing the results of AEDA and EDA with the original sample data reveals that AEDA has a greater effect on improving model performance, with improvements exceeding 10% in all cases.	cache/fcis-30522.pdf	txt/fcis-30522.txt