Skip to Main Content
Table 4 
Highest results for each DSM, using BF1 for Farahmand data set, Pearson r for Reddy (r), and Spearman ρ for all the other data sets. For English, in each pair of values, the first is for the compounds found in the corpus, and the second uses fallback for missing compounds.
Data setPPMI–SVDPPMI–TopKPPMI–threshglovelexvecw2v–cboww2v–sg
Farahmand .487/.424 .435/.376 .472/.404 .400/.358 .449/.431 .512/.471 .507/.468 
  
Reddy (r.738/.726 .732/.717 .762/.768 .783/.787 .787/.787 .803/.798 .814/.814 
  
Reddy (ρ) .743/.743 .706/.716 .791/.803 .754/.759 .774/.773 .796/.796 .812/.812 
EN-comp .655/.666 .624/.632 .688/.704 .638/.651 .646/.658 .716/.730 .726/.741 
FR-comp .584 .550 .702 .680 .677 .652 .653 
PT-comp .530 .519 .602 .555 .570 .588 .586 
Data setPPMI–SVDPPMI–TopKPPMI–threshglovelexvecw2v–cboww2v–sg
Farahmand .487/.424 .435/.376 .472/.404 .400/.358 .449/.431 .512/.471 .507/.468 
  
Reddy (r.738/.726 .732/.717 .762/.768 .783/.787 .787/.787 .803/.798 .814/.814 
  
Reddy (ρ) .743/.743 .706/.716 .791/.803 .754/.759 .774/.773 .796/.796 .812/.812 
EN-comp .655/.666 .624/.632 .688/.704 .638/.651 .646/.658 .716/.730 .726/.741 
FR-comp .584 .550 .702 .680 .677 .652 .653 
PT-comp .530 .519 .602 .555 .570 .588 .586 
Close Modal

or Create an Account

Close Modal
Close Modal