Table 2:

Performance of models under different promptings and data integration methods on string-level metrics. ↑ indicates the higher the better.

Model : Parameters	Prompting	Method	TSR (%)↑	Levenshtein ↑	ExactMatch ↑
qwen2:7b	FS	UGP	25.107	0.043	0.001
qwen2:7b	ZS	UGP	27.748	0.057	0.001
qwen2:7b	ZS	FCSP	43.477	0.297	0.211
qwen2:7b	ZS	UCP	43.513	0.298	0.210
qwen2:7b	FS	UCP	48.592	0.343	0.259
qwen2:7b	FS	FCSP	48.693	0.344	0.260
qwen2.5:7b	ZS	UGP	28.432	0.071	0.001
qwen2.5:7b	FS	UGP	28.627	0.085	0.031
qwen2.5:7b	ZS	FCSP	51.946	0.423	0.345
qwen2.5:7b	ZS	UCP	52.041	0.425	0.345
qwen2.5:7b	FS	FCSP	56.377	0.453	0.383
qwen2.5:7b	FS	UCP	56.456	0.454	0.383
qwen2.5:32b	FS	UGP	23.378	0.032	0.001
qwen2.5:32b	ZS	UGP	28.805	0.082	0.001
qwen2.5:32b	ZSCOT	UCP	49.680	0.385	0.284
qwen2.5:32b	ZS	UCP	49.917	0.386	0.285
qwen2.5:32b	ZS	FCSP	50.203	0.387	0.288
qwen2.5:32b	ZSCOT	FCSP	50.203	0.387	0.288
qwen2.5:32b	FS	UCP	51.332	0.387	0.292
qwen2.5:32b	FS	FCSP	51.467	0.388	0.296
llama3:8b	FS	UGP	26.072	0.047	0.001
llama3:8b	ZS	UGP	28.732	0.066	0.001
llama3:8b	ZS	FCSP	49.462	0.343	0.259
llama3:8b	ZS	UCP	49.536	0.344	0.258
llama3:8b	FS	UCP	50.255	0.362	0.283
llama3:8b	FS	FCSP	50.351	0.363	0.285
gPt4o	FS	UGP	23.416	0.033	0.001
gPt4o	ZS	UGP	23.452	0.033	0.001
gPt4o	FS	UCP	30.862	0.136	0.046
gPt4o	FS	FCSP	31.051	0.132	0.044
gPt4o	ZS	FCSP	31.107	0.132	0.044
gpt4o	ZS	UCP	31.148	0.135	0.046