Skip to Main Content
Table 20 

Performance of UKB on Nat-Nat configuration when injected with the pseudoword-specific (specific) and average sense distribution (average) information as well as the original performance values for UKB and IMS (as reported earlier in Table 18). All the improvements of the UKB (specific) and UKB (average) systems over the UKB (original) system are statistically significant at the p < 0.001 level.

Size of training data080160240320400480560640720800
IMS 81.9* 84.4 86.4 87.4 88.2 88.7 89.3 89.7 90.0 90.3 
UKB (original) 38.8 56.5 59.2 61.1 62.1 62.8 63.4 63.8 64.2 64.4 64.6 
UKB (specific) 65.6 78.0* 78.8 79.2 79.5 79.7 79.9 80.1 80.3 80.4 80.4 
UKB (average) 58.7 75.2 75.8 76.3 76.7 76.8 76.9 77.0 77.1 77.2 77.2 
Size of training data080160240320400480560640720800
IMS 81.9* 84.4 86.4 87.4 88.2 88.7 89.3 89.7 90.0 90.3 
UKB (original) 38.8 56.5 59.2 61.1 62.1 62.8 63.4 63.8 64.2 64.4 64.6 
UKB (specific) 65.6 78.0* 78.8 79.2 79.5 79.7 79.9 80.1 80.3 80.4 80.4 
UKB (average) 58.7 75.2 75.8 76.3 76.7 76.8 76.9 77.0 77.1 77.2 77.2 
Close Modal

or Create an Account

Close Modal
Close Modal