Naslov
Predicting Cyberbullying using Machine Learning

Autori
Zaninović, Ana

Vrsta, podvrsta i kategorija rada
Ocjenski radovi, diplomski rad, preddiplomski

Fakultet
Fakultet elektrotehnike i računarstva

Mjesto
Zagreb

Datum
26.06

Godina
2023

Stranica
32

Mentor
Bagić Babac, Marina

Ključne riječi
machine learning ; cyberbullying ; sentiment ; classification

Sažetak
This thesis aimed to analyze social media data and recognize whether or not it contains bullying. First, emotion analysis of used datasets was made using NRC Emotion Intensity Lexicon (NRC EIL) for detecting emotion distribution. The used datasets consisted of text samples collected from various social media platforms, and corresponding labels which indicated present of different factors including bullying, aggression, attack, toxicity, and bullying on the basis of age, gender, religion, ethnicity and other factors. Preprocessing techniques like tokenization, stemming, and stop-word removal were applied to clean the text data. Then, machine learning algorithms like Naive Bayes, Multinomial NM, Bernoulli NB, Logistic Regression, SGDC, SVC, Linear SVC and NuSVC were utilized for text classification.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo, Informacijske i komunikacijske znanosti

POVEZANOST RADA