Skip to main content

Table 7 Relations extracted by each sieve on development and test datasets.

From: Sieve-based relation extraction of gene regulatory networks from biological literature

 

Dev

Test

Sieve

#

SER

#

SER

Event extraction

29

1.00

32

1.00

Mention processing

44

0.87

12

1.00

Event processing

11

0.84

2

1.00

Gene processing

14

0.84

5

0.97

Event-based gene processing1

26

0.73

15

0.96

Rule-based processing

12

0.75

53

0.76

Data cleaning

22/20

0.75

14/5

0.73

  1. Data cleaning results represent the number of loop relations and the number of redundant relations (separated by forward slash). Slot error rate (SER) results are cumulative.
  2. 1 Due to additional analysis we saw that the event-based gene processing sieve does not improve the final results, therefore we do not employ this sieve on the test data for the final result.
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us