Skip to Main Content
Table 6: 
Comparison of various embedding methods for (a) analogy tests (performance is measured in accuracy) and (b) word semantic similarity tests (performance is measured in terms of Spearman rank correlation).
Embedding(a) Analogy(b) Semantic
Google-SemGoogle-SynMSRRGMTurkRWMENSimLex999AP
GloVe 79.02 52.26 51.49 75.29 64.27 31.63 72.19 34.86 60.70 
Hard-GloVe 80.26 62.76 51.59 76.50 64.26 31.45 72.19 35.17 59.95 
GN-GloVe 76.13 51.00 49.29 74.11 66.36 36.20 74.49 37.12 61.19 
GP-GloVe 79.15 51.55 48.88 75.30 63.46 27.64 69.78 34.02 57.71 
 
RAN-GloVe 80.29 62.89 50.98 76.22 64.09 31.33 72.09 34.36 61.69 
Embedding(a) Analogy(b) Semantic
Google-SemGoogle-SynMSRRGMTurkRWMENSimLex999AP
GloVe 79.02 52.26 51.49 75.29 64.27 31.63 72.19 34.86 60.70 
Hard-GloVe 80.26 62.76 51.59 76.50 64.26 31.45 72.19 35.17 59.95 
GN-GloVe 76.13 51.00 49.29 74.11 66.36 36.20 74.49 37.12 61.19 
GP-GloVe 79.15 51.55 48.88 75.30 63.46 27.64 69.78 34.02 57.71 
 
RAN-GloVe 80.29 62.89 50.98 76.22 64.09 31.33 72.09 34.36 61.69 
Close Modal

or Create an Account

Close Modal
Close Modal