Skip to Main Content
Table 5 

Overview of gold partitionability estimates and of clusterability measures to be evaluated.

Gold partitionability estimates Umid: proportion of mid-range (2–4) instance similarity ratings for a lemma Uiaa: inter-annotator agreement on the Usim data set (average pairwise Spearman) 
Intra-clust clusterability measures vr, wpr, sep based on k-means clustering k estimated as comps clustering computed based on either lexsub or clls substitutes 
Inter-clust clusterability measures comparing comps partitioning of clls with comps partitioning of lexsub comparison either through V or pF 
Baseline average number ncs of cliques clusters, computed either from lexsub or clls data 
Gold partitionability estimates Umid: proportion of mid-range (2–4) instance similarity ratings for a lemma Uiaa: inter-annotator agreement on the Usim data set (average pairwise Spearman) 
Intra-clust clusterability measures vr, wpr, sep based on k-means clustering k estimated as comps clustering computed based on either lexsub or clls substitutes 
Inter-clust clusterability measures comparing comps partitioning of clls with comps partitioning of lexsub comparison either through V or pF 
Baseline average number ncs of cliques clusters, computed either from lexsub or clls data 
Close Modal

or Create an Account

Close Modal
Close Modal