Klasifikacija sentimenta i emocija u javnim komentarima iz domene telekoma

undergraduate thesis

• Title (english): Sentiment and Emotion Classification in Public Comments from the Telecom Domain
• Author: Vedran Kurdija
• Mentor: Jan Šnajder (mentor)
• Committee member: Jan Šnajder (predsjednik povjerenstva)
• Committee member: Martin Tutek (član povjerenstva)
• Committee member: Josip Jukić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Komentari korisnika sadrže neprocjenjive informacije za pružatelje usluga. Značajan dio tih informacija leži u emocijama i sentimentu komentara. Ručna obrada tih komentara skupa je i vremenski zahtjevna. Cilj ovog rada bio je osmisliti i implementirati modele dubokog učenja koji mogu obavljati ovaj zadatak.
Provedena je analiza skupa podataka CATACX koji sadrži javne korisničke komentare na hrvatskom jeziku s označenim emocijama i sentimentima. Istražene su korelacije između sentimenata i emocija. Razvijeni su i implementirani modeli strojnog i dubokog učenja: logistička regresija, jedinica za dugotrajno kratkoročno pamćenje (LSTM) i dvosmjerni LSTM. Modeli su upotrijebljeni za klasifikaciju emocija i sentimenata, a ostvarili su značajno bolje rezultate u predviđanju sentimenata, što ukazuje na to da je predviđanje emocija složeniji zadatak.

Abstract (english)

User comments contain invaluable information for service providers. A significant part of this information lies in the emotions and sentiment of the comment. Manual processing of these comments is expensive and time-consuming. The goal of this thesis was to develop and implement deep learning models that can perform this task.
An analysis of the CATACX dataset containing public user comments in Croatian with labeled emotions and sentiments was performed, and correlations between sentiments and emotions were investigated. Machine learning and deep learning models were developed and implemented, a logistic regression model, a long short-term memory (LSTM) model, and a bidirectional LSTM model. The models were used to classify emotions and sentiments and performed considerably better in predicting sentiments, suggesting that predicting emotions is a more difficult task.

• Keywords: klasifikacija sentimenta; klasifikacija emocija; obrada prirodnog jezika; strojno učenje; duboko učenje; LSTM
• Keywords (english): sentiment classification; emotion classification; natural language processing; machine learning; deep learning; LSTM
• Language: english
• URN:NBN: urn:nbn:hr:168:492047
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 09:14:56