Table 3 Performance of lexical features on named entity classification. F1: bag-of-n-grams; F2: boundary n-grams; F3: sliding character window; F4: boundary substrings; F5: morphology patterns.
From: Incorporating rich background knowledge for gene named entity classification and recognition
Feature | All terms (P/R/F1) | OOV terms (P/R/F1) |
|---|---|---|
F1 | 53.97/70.72/61.22 | -- |
F1+F2 | 73.98/67.13/70.39 | -- |
F1+F2+F3 | 72.54/71.15/71.84 | 50.88/2.94/5.56 |
F1+F2+F3+F4 | 73.75/85.22/79.07 | 70.93/75.89/73.32 |
F1+F2+F3+F5 | 72.81/86.46/79.05 | 71.00/74.16/72.55 |
F1+F2+F3+F4+F5 | 75.52/85.63/80.26 | 74.03/75.68/74.85 |