File(s) not publicly available
High precision extraction of grammatical relations
presentationposted on 2023-06-08, 07:44 authored by John Carroll, Ted Briscoe
A parsing system returning analyses in the form of sets of grammatical relations can obtain high precision if it hypothesises a particular relation only when it is certain that the relation is correct. We operationalise this technique---in a statistical parser using a manually-developed wide-coverage grammar of English---by only returning relations that form part of all analyses licensed by the grammar. We observe an increase in precision from 75% to over 90% (at the cost of a reduction in recall) on a test corpus of naturally-occurring text.
PublisherMorgan Kaufmann Publishers Inc
Event nameProceedings of the 19th International Conference on Computational Linguistics (COLING'02)
Event locationTaipei, Taiwan
Department affiliated with
- Informatics Publications
NotesOriginality: Presents a generally-applicable technique for increasing the precision of the analyses returned by natural language parsers by considering the relative weighting assigned by the parser to all possible sub-analyses. Rigour: Technique tested on a publicly-available domain-independent gold standard test set. Significance: Results show up to a 15 percentage point increase in precision (at a cost in recall). Impact: Google Scholar 33 citations. Outlet: Presented at a long-established, top-tier international conference on natural language processing.
Full text available