Umeå universitets logga
umu.se
Publikationer
Vänta ...
Enkel sökning
Avancerad sökning -
Forskningspublikationer
Avancerad sökning -
Studentuppsatser
Statistik
English
Svenska
Norsk
Ändra sökning
Sök
Exportera
JSON SweCris
Länk till posten
Permanent länk
Direktlänk
http://umu.diva-portal.org/smash/project.jsf?pid=project:1146
BETA
Projekt
Projekttyp/Bidragsform
Projektbidrag
Titel [sv]
Parametriserad syntaktisk analys för naturliga språk
Titel [en]
Parameterized Natural Language Parsing
Abstract [sv]
Language technology, including natural language processing, is a wide area of great scientific and commercial interest. Parsing, i.e., analysis of syntactic structure, is still a bottleneck in many such applications.We propose a detailed study of the time complexity of natural language parsing problems. A number of different grammatical frameworks have been suggested in the literature. We wish to study them in order to find out which aspects of such frameworks make parsing too expensive and under what conditions parsing is tractable. The complexity of parsing problems has been studied before, but the classifications these studies have yielded is generally to coarse. We want to study in more detail how various aspects of the grammatical frameworks contribute to the parsing complexity.As a main tool in our study, we will employ the theory of parameterized complexity, which is ideally suited for isolating the factors that contribute the most to the complexity. Once we have investigated under which conditions parsing is tractable, we want to turn our results into concrete algorithms that can be evaluated empirically. The goal is to help developers of natural language processing applications to choose the tools best suited for the specific needs of their applications.
Projektledare
Björklund, Henrik
Umeå universitet
Koordinerande organisation
Umeå universitet
Forskningsfinansiär
Vetenskapsrådet
Tidsperiod
2012-01-01 - 2015-12-31
Nationell ämneskategori
Datavetenskap (datalogi)
Språkteknologi (språkvetenskaplig databehandling)
Identifikatorer
DiVA, id: project:1146
Projekt id: 2011-06080_VR
Sök vidare i DiVA
I ämnet
Datavetenskap (datalogi)
Språkteknologi (språkvetenskaplig databehandling)
Sök vidare utanför DiVA
Google
Google Scholar
v. 2.44.0
|
WCAG
|
Umeå universitetsbibliotek
|
Registrera i DiVA
|
Support
|
Sök i DiVA portal
DiVA
Logotyp