Table 6.

Performance of TrOCR-ctx across different regions in the UoS_Data_Rescue dataset, providing a logbook-wise analysis to evaluate how the model performs on various logbook types and regions

	Table structure recognition			Tabular data reconstruction
Regions	P	R	F1	Rouge-L	WER	CER	EM	F1 (Char)	F1 (Token)
Tromelin$_{}^{✓}$	0.903	0.903	0.897	0.928	0.088	0.081	0.912	0.949	0.912
Diego-Suarez$_{}^{✓}$	0.836	0.866	0.840	0.902	0.115	0.109	0.885	0.936	0.885
UK$_{}^{✓}$	0.666	0.998	0.782	0.939	0.131	0.121	0.869	0.942	0.869
Ambanja Août-décembre$_{}^{✓}$	0.588	0.944	0.719	0.922	0.131	0.068	0.869	0.945	0.869
South Africa$_{}^{✓}$	0.640	0.954	0.764	0.909	0.133	0.178	0.867	0.895	0.867
Natal, Africa	0.757	0.697	0.720	0.951	0.135	0.117	0.865	0.931	0.865
Tunisia$_{}^{✓}$	0.853	0.935	0.890	0.906	0.153	0.215	0.847	0.904	0.847
Zanzibar$_{}^{✓}$	0.912	0.946	0.928	0.905	0.156	0.162	0.844	0.884	0.844
Algeria$_{}^{✓}$	0.787	0.904	0.832	0.913	0.175	0.135	0.825	0.905	0.825
Tennessee	0.414	0.914	0.530	0.822	0.188	0.218	0.812	0.860	0.812
Libya	0.782	0.881	0.819	0.898	0.199	0.205	0.801	0.858	0.801
Bear	0.352	0.875	0.488	0.834	0.201	0.177	0.799	0.869	0.799
Devon, UK	0.907	0.986	0.943	0.839	0.202	0.182	0.798	0.861	0.798
Arctic$_{}^{✓}$	0.737	0.969	0.834	0.663	0.313	0.327	0.687	0.727	0.687
Egypt$_{}^{\ast }$ $_{}^{✓}$	0.420	0.896	0.552	0.791	0.339	0.260	0.661	0.800	0.661
India$_{}^{\ast }$	0.781	0.915	0.841	0.663	0.345	0.297	0.655	0.789	0.655
Uganda	0.838	0.893	0.864	0.856	0.362	0.460	0.638	0.693	0.638
Morocco	0.508	0.722	0.589	0.706	0.377	0.481	0.623	0.710	0.623
Egypt$_{}^{+}$	0.779	0.822	0.783	0.854	0.423	0.381	0.577	0.744	0.577
Madagascar	0.819	0.923	0.864	0.778	0.426	0.598	0.574	0.627	0.574
Mozambique	0.729	0.765	0.730	0.745	0.469	0.770	0.531	0.603	0.531
India$_{}^{+}$	0.884	0.876	0.879	0.566	0.473	0.576	0.527	0.665	0.527
UK and World$_{}^{✓}$	0.821	0.863	0.834	0.547	0.489	0.812	0.511	0.633	0.511
Mauritius	0.846	0.831	0.834	0.786	0.506	0.725	0.494	0.583	0.494
Ben Nevis, UK	0.981	0.845	0.907	0.627	0.520	0.510	0.480	0.668	0.480
Philippines	0.869	0.858	0.842	0.406	0.711	0.652	0.289	0.482	0.289

$_{}^{\ast }$ The source of these logbooks is from the UK Met Office

$_{}^{+}$ The source of these logbooks are from the US NOAA

$_{}^{✓}$ The logbooks contain a mix of handwritten and typed text