Benchmaking plos

Header metadata

Evaluation on 1000 random PDF files out of 998 PDF (ratio 1.0).

Strict Matching (exact matches)

Field-level results

label	precision	recall	f1	support
authors	99.07	98.97	99.02	969
first_author	99.28	99.17	99.23	969
title	95.54	94.2	94.86	1000
all fields (micro avg.)	97.95	97.41	97.68	2938
all fields (macro avg.)	97.96	97.45	97.7	2938

Soft Matching (ignoring punctuation, case and space characters mismatches)

Field-level results

label	precision	recall	f1	support
authors	99.07	98.97	99.02	969
first_author	99.28	99.17	99.23	969
title	99.09	97.7	98.39	1000
all fields (micro avg.)	99.14	98.6	98.87	2938
all fields (macro avg.)	99.14	98.61	98.88	2938

Levenshtein Matching (Minimum Levenshtein distance at 0.8)

Field-level results

label	precision	recall	f1	support
authors	99.48	99.38	99.43	969
first_author	99.38	99.28	99.33	969
title	99.7	98.3	98.99	1000
all fields (micro avg.)	99.52	98.98	99.25	2938
all fields (macro avg.)	99.52	98.99	99.25	2938

Ratcliff/Obershelp Matching (Minimum Ratcliff/Obershelp similarity at 0.95)

Field-level results

label	precision	recall	f1	support
authors	99.38	99.28	99.33	969
first_author	99.28	99.17	99.23	969
title	99.29	97.9	98.59	1000
all fields (micro avg.)	99.32	98.77	99.04	2938
all fields (macro avg.)	99.32	98.78	99.05	2938

Instance-level results

Total expected instances:   1000
Total correct instances:    937 (strict)
Total correct instances:    971 (soft)
Total correct instances:    978 (Levenshtein)
Total correct instances:    974 (ObservedRatcliffObershelp)

Instance-level recall:  93.7    (strict)
Instance-level recall:  97.1    (soft)
Instance-level recall:  97.8    (Levenshtein)
Instance-level recall:  97.4    (RatcliffObershelp)

Evaluation metrics produced in 3.302 seconds