Automated annotation of chemical names in the literature with tunable accuracy

Table 4 The comparison of precision (P), recall (R), and F measure (F) of this work (MAA) with those of other studies.

	Corpus	Dictionaries	Filters	P	R	F
Hettne	Kolarik's Corpus	MeSH C MeSH S	Term Disambiguation	0.75	0.22	0.34
			Term disambiguation	0.83	0.07	0.13
Kolarik	Kolarik's Corpus	MeSH C MeSH S	Free Text	0.34	0.27	0.30
			Free Text	0.44	0.10	0.16
MAA (this work)	Kolarik's Corpus	MeSH C+S	Free Text	0.44	0.37	0.41
			MeSH Filter	0.61	0.34	0.43
			Token & Rule	0.66	0.33	0.43
			Protein & Gene	0.65	0.32	0.43
			TP Filter 0.025	0.73	0.28	0.41
			TP Filter 0.15	0.76	0.27	0.40
			TP Filter 0.30	0.79	0.26	0.39
			TP Filter 0.40	0.77	0.23	0.35
NLM's MTI*		All MeSH		0.29	0.55	0.38
MAA (this work)	A and T	MeSH C+S	TP Filter 0.15	0.68	0.65	0.66
	1S, LS and T		TP Filter 0.15	0.84	0.40	0.55
	T		TP Filter 0.15	0.95	0.30	0.46

The filters of MAA are applied cumulatively from "Free Text" to "TP Filter 0.4".
*See reference [20], this work use 273 articles with full text
MTI: Medical Text Indexer
MeSH C: MeSH tree chemical terms
MeSH S: MeSH substances
A: Abstract
T: Title
1S: First sentence of abstract
LS: Last sentence of abstract

ISSN: 1758-2946