Pregled bibliografske jedinice broj: 779266
Getting the Agenda Right: Measuring Media Agenda using Topic Models
Getting the Agenda Right: Measuring Media Agenda using Topic Models // Proceedings of the 2015 Workshop on Topic Models: Post-Processing and Applications
Melbourne, Australija: The Association for Computing Machinery (ACM), 2015. str. 61-66 (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), znanstveni)
CROSBI ID: 779266 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
Getting the Agenda Right: Measuring Media Agenda using Topic Models
Autori
Korenčić, Damir ; Ristov, Strahil ; Šnajder, Jan
Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni
Izvornik
Proceedings of the 2015 Workshop on Topic Models: Post-Processing and Applications
/ - : The Association for Computing Machinery (ACM), 2015, 61-66
ISBN
978-1-4503-3784-7
Skup
2015 Workshop on Topic Models: Post-Processing and Applications
Mjesto i datum
Melbourne, Australija, 19.10.2015
Vrsta sudjelovanja
Predavanje
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
Agenda setting; Agenda measuring; News media analysis; Topic modeling; Document tagging; Multilabel classification
Sažetak
Agenda setting is the theory of how issue salience is transferred from the media to media audience. An agenda-setting study requires one to define a set of issues and to measure their salience. We propose a semi-supervised approach based on topic modeling for exploring a news corpus and measuring the media agenda by tagging news articles with issues. The approach relies on an off-the-shelf Latent Dirichlet Allocation topic model, manual labeling of topics, and topic model customization. In preliminary evaluation, the tagger achieves a micro F1-score of 0.85 and outperforms the supervised baselines, suggesting that it could be successfully used for agenda-setting studies.
Izvorni jezik
Engleski
Znanstvena područja
Računarstvo
POVEZANOST RADA
Projekti:
HRZZ-IP-2013-11-9623 - Postupci strojnog učenja za dubinsku analizu složenih struktura podataka (DescriptiveInduction) (Gamberger, Dragan, HRZZ - 2013-11) ( CroRIS)
Ustanove:
Fakultet elektrotehnike i računarstva, Zagreb,
Institut "Ruđer Bošković", Zagreb