Skip to main content

Table 9 Estimated recalls for Linguistic Patterns 1, 2 and 3 without SVM classification

From: Identifying named entities from PubMed® for enriching semantic categories

Headwords

Pattern 1

Pattern 2

Pattern 3

Total

Gene

17.4%

0.8%

3.1%

18.2%

Protein

11.6%

2.3%

5.4%

14.2%

Disease

1.4%

0.6%

6.1%

6.8%

Cell

8.0%

1.0%

3.5%

10.7%

Cells

29.7%

1.7%

2.7%

31.6%

Average

13.6%

1.3%

4.2%

16.3%

  1. As no true labels are available for PubMed terms, recalls were evaluated based on number of SemCat terms occurring in PubMed that were discovered by the pattern.