 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
| l |
Contextual
classification
|
|
|
|
¡ |
Predications of
step 1 provides context
|
|
|
|
l |
e.g. title lines,
author line, affiliation line…
|
|
|
|
¡ |
Context among
lines can be used as features
|
|
|
l |
e.g. title lines
followed by author line
|
|
|
|
¡ |
Combining
contextual features and other
|
|
|
features, then
trained iteratively
|
|