Table 5.
Source | Feature | Algorithm | Precision | Recall | F1-score |
---|---|---|---|---|---|
EHR | W | Glove | 71.97 | 60.33 | 65.64 |
word2vec | 72.48 | 69.34 | 70.87 | ||
FastText | 73.82 | 69.84 | 71.77 | ||
L | Glove | 74.22 | 67.21 | 70.54 | |
word2vec | 80.18 | 70.90 | 75.25 | ||
FastText | 76.71 | 66.86 | 71.44 | ||
GenMed | W | Glove | 70.42 | 64.73 | 67.46 |
word2vec | 75.15 | 68.91 | 71.90 | ||
FastText | 75.02 | 66.08 | 70.26 | ||
L | Glove | 73.53 | 65.86 | 69.49 | |
word2vec | 75.15 | 63.31 | 68.72 | ||
FastText | 72.18 | 66.64 | 69.30 | ||
Gen | W | Glove | 71.54 | 65.65 | 68.47 |
word2vec | 70.63 | 61.60 | 65.81 | ||
FastText | 73.97 | 62.74 | 67.90 | ||
L | Glove | 73.00 | 64.66 | 68.57 | |
word2vec | 74.11 | 63.38 | 68.32 | ||
FastText | 75.87 | 63.38 | 69.06 | ||
EHR | W,L | word2vec | 76.00 | 69.00 | 72.27 |
W+L | word2vec | 76.52 | 65.93 | 70.83 | |
W-L | word2vec | 75.18 | 67.71 | 71.25 |
Embeddings of base-units on top and, below, with the base-units combined