Skip to Main Content
Table 2 

Mean squared error of different models' predictions, trained on PMI (left) vs. PPMI vectors (right).

modelPMIPPMI
allposnegnear-0allposnegnear-0
additive 0.75 3.11 5.86 (≈0.00) 0.71 1.96 5.88 (0.02) 
multiplicative 0.61 2.50 3.38 0.55 0.59 5.92 3.40 0.62 
weighted additive 0.39 2.52 0.41 0.35 0.32 3.01 0.62 0.27 
full additive 0.56 3.02 0.68 0.54 0.34 1.93 0.63 0.29 
lexical function 0.73 2.92 0.74 0.68 0.45 2.01 0.68 0.37 
shifted additive 0.66 4.66 3.01 0.39 0.48 2.16 3.52 0.18 
modelPMIPPMI
allposnegnear-0allposnegnear-0
additive 0.75 3.11 5.86 (≈0.00) 0.71 1.96 5.88 (0.02) 
multiplicative 0.61 2.50 3.38 0.55 0.59 5.92 3.40 0.62 
weighted additive 0.39 2.52 0.41 0.35 0.32 3.01 0.62 0.27 
full additive 0.56 3.02 0.68 0.54 0.34 1.93 0.63 0.29 
lexical function 0.73 2.92 0.74 0.68 0.45 2.01 0.68 0.37 
shifted additive 0.66 4.66 3.01 0.39 0.48 2.16 3.52 0.18 
Close Modal

or Create an Account

Close Modal
Close Modal