E09-1072 annotation . First of all , the automatic annotation is completely consistent at the
E06-2013 Abstract We describe a system for automatic annotation of English text in the FrameNet
D08-1055 test phase . The accuracy of the automatic annotation is about 90 % . 4.2 Baseline
D13-1042 not possible to verify all the automatic annotations manually . Based on a small-scale
D08-1001 potential of parallel corpora for automatic annotation of data . Using manually edited
E09-1026 semantic role labeling show that the automatic annotations produced by our method improve
A00-3004 usage . Therefore , a complete automatic annotation system based on the approach
E09-1026 relative quality of manual and automatic annotation . Expanding a seed corpus with
C02-2008 gradually increase the accuracy of automatic annotation . 3.2 Word Sense Annotation Inthecomputational
C02-1098 gradually increase the accuracy of automatic annotation . In principle , the tag set
D12-1038 first describe the technology of automatic annotation transformation , which is based
E06-2013 sponsored research project . <title> Automatic Annotation for All Semantic Layers in FrameNet
E03-1072 project member ) corrected the automatic annotations . The annotators had a reference
A97-2016 human annota - tor , manual and automatic annotation are combined in an interactive
E09-1072 postmodifiers in PP-attachment . The automatic annotation thus captures a purely linguistic
C04-1078 two ways . First , we repeat the automatic annotation process until it satisfies the
D12-1038 first describes the technology of automatic annotation transformation , then introduces
E03-1068 resulting from a combination of automatic annotation and manual post-editing . A case
D10-1017 decode ) produces a new set of automatic annotations that can be combined with the
D12-1013 most informative MA samples for automatic annotation . Empirical studies show that
hide detail