Normalization
Normalization is commonly used in NLP to abstract away lexical variations (such as case, punctuation, spelling variants, inflectional variants, etc.) of words with same/similar meaning to increase the recall rate. Different project might have different normalization according to the requirements. STMT includes three normalization applying lexical tools APIs and are described as follows:
Comparison
| Normalize | Operation | Usage | |
|---|---|---|---|
| LexItem Norm |
| one to one | Find term in Lexicon (lsf) |
| Synonym Norm |
| one to many | Find synonym of a term (smt) |
| Lvg Norm |
| one to many | Term to CUI mapping (smt) |