Traditional Chinese Medicine Text Similarity Calculation Model Based on the Bidirectional Temporal Siamese Network

Table 4

The comparison of different similarity calculation models.

Datasets	Financial data		TCM data
Models	Training set accuracy	Test set accuracy	Training set accuracy	Test set accuracy
TF-IDF	0.7042	0.7102	0.5967	0.5510
Word2vec	0.7242	0.6954	0.6441	0.6013
Siamese-LSTM	0.8156	0.7886	0.6574	0.6103
Siamese-CNN	0.8273	0.8097	0.7517	0.7502
Siamese-RNN	0.8039	0.7830	0.7383	0.7129
BTSN	0.8328	0.8311	0.7967	0.7827

Bold values indicate the best record in the comparative test.