Skip to main content

Table 3 Hierarchical SVM classifier performance on Diabetes training dataset for Level 3 predictions.

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

Level 3 category

Curatable abstracts

AUC individual category (SVM)

Category prediction accuracy (%)(MLP)

GAD

148

.966

87.2

HSP

19

.997

89.5

IA2

26

.987

80.8

IGRP

16

.995

75.0

INSULIN

126

.964

87.3

OTH

79

.823

53.2

VAR

29

.695

13.8

Total

443

 

75.6

  1. The AUC and category prediction accuracy values are entered for each Diabetes Level 3 category.