| dc.contributor.advisor | Gulla, Jon Atle | nb_NO |
| dc.contributor.author | Krakeli, Marius | nb_NO |
| dc.contributor.author | Midtsund, Martin Akre | nb_NO |
| dc.date.accessioned | 2014-12-19T13:40:30Z | |
| dc.date.available | 2014-12-19T13:40:30Z | |
| dc.date.created | 2013-10-15 | nb_NO |
| dc.date.issued | 2013 | nb_NO |
| dc.identifier | 656463 | nb_NO |
| dc.identifier | ntnudaim:7716 | nb_NO |
| dc.identifier.uri | http://hdl.handle.net/11250/253459 | |
| dc.description.abstract | Locations within news recommendation is a field with room for improvement. Location is not used by popular news recommendation applications on the market today, and while the location technology has evolved, the research around it regarding news, seems to be incomplete. The use of physical location together with technology has expanded within the latest years, and the demand for it to become part of advanced news recommendation is present. This paper will illuminate what is needed from an enhanced location API to be durable enough to become part of a news recommender system. By implementing an actual location module for a real news recommendation system, we will test different toponym resolution strategies, to get more accurate location data for locations present in news articles. Results show that already existing location services are quite good, and that it is possible to improve their performance by implementing an enhanced layer on top of it with even more intelligence. | nb_NO |
| dc.language | eng | nb_NO |
| dc.publisher | Institutt for datateknikk og informasjonsvitenskap | nb_NO |
| dc.title | Enhanced Location API | nb_NO |
| dc.type | Master thesis | nb_NO |
| dc.source.pagenumber | 96 | nb_NO |
| dc.contributor.department | Norges teknisk-naturvitenskapelige universitet, Fakultet for informasjonsteknologi, matematikk og elektroteknikk, Institutt for datateknikk og informasjonsvitenskap | nb_NO |
