Putting the t where it belongs: Solving a confusion problem in Dutch
A common Dutch writing error is to confuse a word ending in -d with a neighbor word ending in -dt. In this paper we describe the development of a machine-learning-based disambiguator that can determine which word ending is appropriate, on the basis of its local context. We develop alternative disambiguators, varying between a single monolithic classifier and having multiple confusable experts disambiguate between confusable pairs. Disambiguation accuracy of the best developed disambiguators exceeds 99%; when we apply these disambiguators to an external test set of collected errors, our detectio... Mehr ...
Verfasser: | |
---|---|
Dokumenttyp: | conferenceObject |
Erscheinungsdatum: | 2008 |
Sprache: | Englisch |
Permalink: | https://search.fid-benelux.de/Record/base-26646552 |
Datenquelle: | BASE; Originalkatalog |
Powered By: | BASE |
Link(s) : | http://hdl.handle.net/11858/00-001M-0000-0012-3E76-3 |
A common Dutch writing error is to confuse a word ending in -d with a neighbor word ending in -dt. In this paper we describe the development of a machine-learning-based disambiguator that can determine which word ending is appropriate, on the basis of its local context. We develop alternative disambiguators, varying between a single monolithic classifier and having multiple confusable experts disambiguate between confusable pairs. Disambiguation accuracy of the best developed disambiguators exceeds 99%; when we apply these disambiguators to an external test set of collected errors, our detection strategy correctly identifies up to 79% of the errors.