diplomsko delo
Jan Bajt (Author), Marko Robnik Šikonja (Mentor)

Abstract

V diplomskem delu primerjamo slovenske medije s pomočjo analize tematik in sentimenta člankov. Želeli smo analizirati različna stališča medijev do specifičnih političnih dogodkov oziroma tematik. Tematike smo modelirali z modelom LDA, s katerim smo v množici slovenskih člankov poiskali tiste s politično vsebino. Za nalogo zaznavanja sentimenta smo prilagodili model SloBERTa in ga uporabili pri klasifikaciji izbranih člankov v eno izmed treh oznak (pozitivno, nevtralno, negativno). Primerjavo medijev izvedemo na nekaj različnih političnih temah, kjer opazimo nekaj razlik med skupinami medijev. Rezultate predstavimo in izpostavimo nekaj slabosti našega sistema ter podamo predloge za izboljšavo.

Keywords

obdelava naravnega jezika;model BERT;latentna Dirichletova alokacija;modeliranje tematik;detekcija sentimenta;slovenski mediji;računalništvo in informatika;univerzitetni študij;diplomske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UL FRI - Faculty of Computer and Information Science
Publisher: [J. Bajt]
UDC: 004.8:81'322(043.2)
COBISS: 77669123 Link will open in a new window
Views: 369
Downloads: 149
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Topic and sentiment analysis of Slovene media using natural language processing tools
Secondary abstract: We compare topics covered by Slovenian media by analysing sentiment of the articles. We aim to analyse different stances of media towards specific political events or topics. We used LDA model for topic modeling and based on results, we selected articles with political content. For the sentiment analysis task we fine-tuned Slovenian SloBERTa model which we used to classify articles in one of three sentiment labels (positive, neutral, negative). We compare the media on a few political topics, where we notice differences between media. We present the results, highlight weaknesses of our system and suggest improvements.
Secondary keywords: natural language processing;model BERT;latent Dirichlet allocation;topic modeling;sentiment detection;Slovenian media;computer and information science;diploma;Računalniško jezikoslovje;Mediji;Umetna inteligenca;Računalništvo;Univerzitetna in visokošolska dela;
Type (COBISS): Bachelor thesis/paper
Study programme: 1000468
Embargo end date (OpenAIRE): 1970-01-01
Thesis comment: Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages: 47 str.
ID: 13394699