dc.contributor.advisor | Hetland, Magnus Lie | nb_NO |
dc.contributor.author | Heimark, Karl Johan V | nb_NO |
dc.date.accessioned | 2014-12-19T13:42:04Z | |
dc.date.available | 2014-12-19T13:42:04Z | |
dc.date.created | 2014-11-13 | nb_NO |
dc.date.issued | 2013 | nb_NO |
dc.identifier | 762876 | nb_NO |
dc.identifier | ntnudaim:9691 | nb_NO |
dc.identifier.uri | http://hdl.handle.net/11250/253938 | |
dc.description.abstract | For decades there has been done a lot of work on trying to make computers answernatural language questions. Current research still has a long way to go before accom-plishing this goal. In this thesis I have built a prototype based on Quepy, an open sourcesystem for translating questions in natural language to SPARQL and SQL queries. Thisprototype is designed to answer questions regarding the american football league NFL.This report covers the implementation of this prototype as well as an evaluation of it,and a comparison to the current state of the art system. This prototype can answer allquestions I made about the data it was given. It performed better than the state of theart system on these questions. | nb_NO |
dc.language | eng | nb_NO |
dc.publisher | Institutt for datateknikk og informasjonsvitenskap | nb_NO |
dc.subject | ntnudaim:9691 | no_NO |
dc.subject | MTDT Datateknologi | no_NO |
dc.subject | Komplekse datasystemer | no_NO |
dc.title | A Specialized Question Answering System Using Natural Language: Developed Using the Quepy Framework | nb_NO |
dc.type | Master thesis | nb_NO |
dc.source.pagenumber | 101 | nb_NO |
dc.contributor.department | Norges teknisk-naturvitenskapelige universitet, Fakultet for informasjonsteknologi, matematikk og elektroteknikk, Institutt for datateknikk og informasjonsvitenskap | nb_NO |