From: CHEMDNER system with mixed conditional random fields and multi-scale word clustering
Character feature | Wi-4, Wi-3, Wi-2, Wi-1, Wi, Wi+1, Wi+2, Wi+3, Wi+4 |
---|---|
2-gram character feature | Wi-1i, Wii+1 |
3-gram character feature | Wi-2i, Wi-1i+1, Wii+2 |
4-gram character feature | Wi-3i, Wi-2i+1, Wi-1i,Wii+1 |
5-gram character feature | Wi-4i, Wi-3i+1, Wi-2i+2, Wi-1i+3, Wii+4 |