Table 2 Parameters for training W2V embeddings
From: Comparison of different feature extraction methods for applicable automated ICD coding
Parameters | Descriptions | For Fuwai dataset | For CodiEsp dataset |
|---|---|---|---|
sg | Whether Skip-gram is used. | 1 | 1 |
size | The dimension of resulting embeddings. | 256 | 128 |
window | The length of a text window. | 5 | 5 |
iter | Training on data for how many iterations. | 5 | 5 |
min_count | Discarding words/characters appearing less than how many times. | 3 | 3 |