| W10-3005 | In the first phase , they use | IGTREE algorithm | to train a classifier with 3 |
| W09-1203 | with Classifier 1 . We use the | IGTREE algorithm | as implemented in TiMBL . The |
| W09-1203 | taken as the sense . We use the | IGTREE algorithm | . Instances represent predicates |
| W99-0629 | structure , we first train the | IGTree algorithm | , and then let it classify the |
| W06-3604 | including a description of the | IGTREE algorithm | central to our study . In Section |
| W00-0720 | score on the tagging task . The | IGTREE algorithm | is more suited for problems where |
| W00-0720 | accuracy when using the B31-IG or | IGTREE algorithm | . These differences are significant |
| N10-1105 | . For known words , we use the | IGTree algorithm | and 2 words to the left , their |
| W00-0720 | irrelevant ) could also explain why the | IGTREE algorithm | seems to benefit less from the |
| W97-1015 | , we have also implemented the | IGTree algorithm | ( Daelemans et al. , 1997 ) , |
| P12-2035 | . For known words , we use the | IGTree algorithm | and 2 words to the left , their |
| W96-0102 | the IB-IG algorithm . 3.1 The | IGTree Algorithms | IGTree combines two algorithms |
