Cost sensitive hierarchical document classification to triage PubMed abstracts for manual curation

BMC Bioinformatics

Table 2 Hierarchical SVM classifier performance on Autoimmunity training dataset for Level 2 predictions.

Level 2 category	Curatable abstracts	AUC individual category (SVM)	Category prediction accuracy (%)(MLP)
Beta-Amyloid	213	.998	97.2
Diabetes	443	.998	96.6
General Autoimmune	1311	.987	87.6
Lupus	704	.988	94.6
Multiple Sclerosis	957	.990	97.3
Myasthenia Gravis	221	.992	95.0
Rheumatoid Arthritis	501	.990	84.0
Total	4350		92.2

Performance was evaluated with five-fold cross-validation. The AUC and category prediction accuracy values are entered for each Autoimmunity Level 2 category.

ISSN: 1471-2105