Skip to main content

Table 2 Comparison with two unsupervised WSD systems.

From: Fast max-margin clustering for unsupervised word sense disambiguation in biomedical texts

Keyword

All-in-1

SenseClusters

SENSATIONAL

ANA

0.63

0.99

1.0

BPD

0.40

0.65

0.53

BSA

0.50

0.99

0.95

CML

0.55

0.99

0.90

cold

0.37

0.63

0.67

culture

0.52

0.55

0.82

discharge

0.66

0.90

0.95

fat

0.51

0.55

0.53

fluid

0.64

0.88

0.99

glucose

0.51

0.69

0.51

inflammation

0.35

0.47

0.50

inhibition

0.50

0.55

0.54

MAS

0.50

1.0

1.0

mole

0.78

0.77

0.96

nutrition

0.39

0.5

0.55

pressure

0.52

0.89

0.86

single

0.50

0.87

0.99

transport

0.51

0.52

0.57

VCR

0.79

0.65

0.64

average

0.53

0.74

0.76

  1. SENSATIONAL outperforms the All-in-1 baseline by an impressive 23% and the SenseClusters system by 2% on a task of disambiguating terms in PubMed abstracts. Differences in performance across all words is statistically significant using the Chi-squared test with Yates' correction (two-tailed test with 1 degree of freedom; between All-in-1 and SenseClusters, Chi-square = 488.671, p < 0.0001; between SenseClusters and SENSATIONAL, Chi-square = 5.388, p = 0.0203).