Table 20 Training parameters for embeddings models built in this work

Number of negatives sampled

20

20

Sampling threshold

6e\(-\)5

6e\(-\)5

Minimum number of word occurrences

10

10

Minimum length of character n-gram

3

3

Maximum length of character n-gram

6

6

Size of word vectors

300

300

Epochs

10

10

Processor

4 Intel Xeon 2.00 GHz, 8 Cores, 16 Logical Processors

4 Intel Xeon 2.00 GHz, 8 Cores, 16 Logical Processors

RAM

32 GB

32 GB

Corpus size

1 GB

1 GB

Training time

4 h

8 h