A Tag Sequence Grammar for Norwegian
The RASP parser for English combines (a) a part-of-speech (PoS) tagger, (b) a linguistic grammar of tag sequences, and (c) a stochastic disambiguation component. The resulting analysis system for natural language achives a good balance of robustness and efficiency, on the one hand, and parsing accuracy, on the other hand. Unlike in purely statistical parsers, RASP makes use of a hand-built grammar of English sentence structure, though not assuming lexical knowledge beyond the (relatively coarse-grained) morpho-syntactic information provided by a PoS tagger. The Oslo-Bergen Tagger (OBT) is a full-scale, robust PoS tagger for Norwegian and will serve as the starting point for this project.
This project aims to write a syntactic grammar of Norwegian, based on PoS tag sequences (much in the RASP spirit). The grammar will be implemented in the DELPH-IN typed feature structure formalism, as implemented in the Linguistic Knowledge Builder (LKB) and Platform for Experimentation with Linguistic Processing Techniques (PET) systems. The project assumed a good understanding of unification-based grammar (in the spirit of Sag, Wasow, and Bender, 2003) and general knowledge of Norwegian syntax. Please contact Stephan Oepen to discuss this project further.
| Type: | 60 ECTS |
| Veileder(e): | Stephan Oepen |
| Tilknyttet forskningsgruppe(r): | LNS |