Analiza sentimenta u muzičkim žanrovima

master's thesis

• Title (english): Sentiment Analysis of Music Genres
• Author: Luka Čagalj
• Mentor: Goran Delač (mentor)
• Committee member: Goran Delač (predsjednik povjerenstva)
• Committee member: Marin Šilić (član povjerenstva)
• Committee member: Klemo Vladimir (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2020-09-18, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Analiza sentimenta ostvarena je na više načina. Različite reprezentacije teksta poput TF-IDF reprezentacije, reprezentacije dobivene primjenom rangirajuće funkcije BM25 i GloVe modela za učenje vektorskih reprezentacija riječi, koristile su se za učenje modela. U izradi sustava koristili su se modeli: logistička regresija, stroj potpornih vektora i LSTM povratna neuronska mreža. Za problem binarne klasifikacije sentimenta dobivena je točnost 78\%, dok je za višeklasni problem dobivena točnost 60\% na skupu za testiranje. Zbog nedostatka označenih podataka korišteno je aktivno učenje modela. Naučeni modeli su korišteni u analizi sentimenta unutar glazbenih žanrova.

Abstract (english)

The analysis of sentiment has been accomplished in a number of ways. Various text representations such as TF-IDF representation, representation obtained by applying the BM25 ranking function, and GloVe model for learning vector word representations were used to train models. Models were used in the design of the system: logistic regression, support vector machine and LSTM recurrent neural network. Accuracy of 78 \% was obtained for binary classification and for a multiclass problem, 60 \% accuracy was obtained on the test set. Active learning was used due to the lack of labeled data. Learned models were used in sentiment analysis within musical genres.

• Keywords: analiza sentimenta; TF-IDF; BM25; GloVe; logistička regresija; stroj potpornih vektora; LSTM povratna neuronska mreža; aktivno učenje
• Keywords (english): sentiment analysis; TF-IDF; BM25; GloVe; logistic regression; support vector machine; LSTM recurrent neural network; active learning
• Language: croatian
• URN:NBN: urn:nbn:hr:168:881000
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 11:59:22