The first step in the development of text mining technology for cancer risk assessment: identifying and organizing scientific evidence in risk assessment literature

Table 6 Number of abstracts, keywords and FScore per class

Class	Abstracts	Keywords	F-Measure
Carcinogenic activity	1023	1157	92.8
Human study/epidemiology	190 (171)	44	77.7
Tumor related	39	28	56.3
Morphological effect on tissue/organ	2	1
Biochemical/cellbiological effects	2	3
Biomarkers	35	14	68.4
Polymorphism	37	32	79.5
Animal study	629 (546)	46	80.2
Study length	156 (3)	3
2-year cancer bioassay	14	9
Short and medium	143	110	45.9
Tumors	186	73	74.3
Preneoplastic lesions	150	121	81.2
Morphological effect on tissue/organ	60	50	46.3
Biochemical/cellbiological effects	135	198	52.1
Biomarker	6	3
Type of animal	452 (388)	166	70.5
Genetically modified animals	73	76	73.5
Cell experiments	319 (313)	28	78.5
Biochemical/cellbiological effects	100	128	58.7
Subcellular systems	2	2
Study on microorganisms	44	22	85.2
Mode of Action	653	316	85.5
Genotoxic	426 (72)	16	89.1
Strand breaks	32	12	77.4
Adducts	174	11	89.8
Chromosomal change	84 (36)	23	68.2
Micronucleus	47	5	85.9
Chromosomal aberration	35	10	68.2
Mutations	145	38	85.4
Other dna mods	100	52	62.0
Non-genotoxic	324 (8)	4	76.3
Reactive oxygen species	54	26	70.5
Cytoxicity	50	7	62.0
DNA repair	29	8	64.2
Hormonal receptor	47	30	61.6
Effects on cell proliferation	113	30	69.6
Effects on cell death	110	10	83.3
Transcriptional, translational, posttranslational modifications	27	22	61.2
Peroxisome proliferation	3	2
Inflammation	15	10
Toxicokinetics	365	269	77.7
Absorption, uptake, distribution, excretion	117	45	69.8
Bioaccumulation/Lipophility	0	0
Metabolism	275 (152)	36	76.4
Activation or deactivation	191	161	74.8
Reactive oxygen species	7	6
Toxicokinetic modeling	31	21	84.6

The first column shows the name of a class in the taxonomy. The second column shows the total number abstracts classified in the class (or its sub-classes). The value in brackets is the number of abstracts classified in the class without taking the sub-classes into account. The third column shows the total number of unique keyword annotations for each class. The count does not include the annotations for sub-classes, except for the three top level classes where the number of all keywords (also those of sub-classes) are included.

ISSN: 1471-2105