| P02-1018 | to compare the performance of | empty node recovery | on parser output with the empty |
| D10-1062 | ( 2002 ) defines a pattern for | empty node recovery | to be a minimally connected tree |
| P02-1018 | proposes an evaluation procedure for | empty node recovery | procedures which is independent |
| P04-1040 | to Earlier Work A procedure for | empty node recovery | was first described in ( Johnson |
| W04-2003 | performance by integrating pars - ing , | empty node recovery | and antecedent finding in a single |
| P02-1018 | which measures the accuracy of | empty node recovery | but not co-indexation , is just |
| P02-1018 | performance by integrating parsing , | empty node recovery | and antecedent finding in a single |
| W12-4213 | al. , 2010 ) . We were inspired | empty node recovery | evaluation by ( Johnson , 2002 |
| P02-1018 | in the training corpus during | empty node recovery | , as explained in the text . |
| P02-1018 | an empty complementizer 0 . 3 | Empty node recovery | evaluation The previous section |
