Skip to main content

Table 3 F-scores under the optimal word similarity threshold

From: A hierarchical method to automatically encode Chinese diagnoses through semantic similarity estimation

Measure

4-digit

3-digit

Flat

Hierarchical

Flat

Hierarchical

HowNet-based

0.8986

0.9108

0.9303

0.9427

Word-based vector

0.8932

0.8958

0.9216

0.9369

Char-based vector

0.8884

0.8919

0.9199

0.9237

Exact matching

0.7933

0.7928

0.8532

0.8564