Skip to Main Content
Table 8 

Results of all distributional similarity measures when tuning K over the development set. We encode the description of the measures presented in Table 2 in the following manner—h = health-care corpus; R = RCV1 corpus; b = binary templates; u = unary templates; L = Lin similarity measure; B = BInc similarity measure; pCt = pair of CUI tuples representation; pC = pair of CUIs representation; Ct = CUI tuple representation; C = CUI representation; Lin & Pantel = similarity lists learned by Lin and Pantel.


Edges
Propositions
Dist. sim. measure
Recall
Precision
F1
Recall
Precision
F1
h-b-B-pCt 52.5 31.6 37.7 69.8 50.0 57.1 
h-b-B-pC 50.5 26.5 30.7 67.1 43.5 50.1 
h-b-B-Ct 10.4 44.5 15.4 39.1 78.9 51.6 
h-b-B-C 7.6 42.9 11.1 37.9 79.8 50.7 
h-b-L-pCt 53.4 34.9 37.5 73.5 50.6 56.1 
h-b-L-pC 47.2 35.2 35.6 68.6 52.9 56.2 
h-b-L-Ct 47.0 26.6 30.2 64.9 47.4 49.6 
h-b-L-C 34.6 22.9 22.5 57.2 52.6 47.6 
h-u-B-Ct 5.1 37.4 8.5 35.1 91.0 49.7 
h-u-B-C 7.2 42.4 11.5 36.1 90.3 50.1 
h-u-L-Ct 22.8 22.0 18.3 49.7 49.2 44.5 
h-u-L-C 16.7 26.3 17.8 47.0 56.8 48.1 
R-b-L-l 49.4 21.8 25.2 72.4 39.0 45.5 
R-u-L-l 24.1 30.0 16.8 47.1 55.2 42.1 
R-u-B-l 9.5 57.1 14.1 37.2 84.0 49.5 
Lin & Pantel 37.1 32.2 25.1 58.9 54.6 48.6 

Edges
Propositions
Dist. sim. measure
Recall
Precision
F1
Recall
Precision
F1
h-b-B-pCt 52.5 31.6 37.7 69.8 50.0 57.1 
h-b-B-pC 50.5 26.5 30.7 67.1 43.5 50.1 
h-b-B-Ct 10.4 44.5 15.4 39.1 78.9 51.6 
h-b-B-C 7.6 42.9 11.1 37.9 79.8 50.7 
h-b-L-pCt 53.4 34.9 37.5 73.5 50.6 56.1 
h-b-L-pC 47.2 35.2 35.6 68.6 52.9 56.2 
h-b-L-Ct 47.0 26.6 30.2 64.9 47.4 49.6 
h-b-L-C 34.6 22.9 22.5 57.2 52.6 47.6 
h-u-B-Ct 5.1 37.4 8.5 35.1 91.0 49.7 
h-u-B-C 7.2 42.4 11.5 36.1 90.3 50.1 
h-u-L-Ct 22.8 22.0 18.3 49.7 49.2 44.5 
h-u-L-C 16.7 26.3 17.8 47.0 56.8 48.1 
R-b-L-l 49.4 21.8 25.2 72.4 39.0 45.5 
R-u-L-l 24.1 30.0 16.8 47.1 55.2 42.1 
R-u-B-l 9.5 57.1 14.1 37.2 84.0 49.5 
Lin & Pantel 37.1 32.2 25.1 58.9 54.6 48.6 
Close Modal

or Create an Account

Close Modal
Close Modal