Cyberbullying Sentiment Analysis with Word2Vec and One-Against-All Support Vector Machine

Depression and social anxiety are the two main negative impacts of cyberbullying. Unfortunately, a survey conducted by UNICEF on 3rd September 2019 showed that 1 in 3 young people 30 countries had been victims Sentiment analysis research will be to detect comment contains Dataset cyberbullying is obtained from Kaggle website, named, Toxic Comment Classification Challenge. The pre-processing process consists 4 stages, namely generalization (convert text into lowercase remove punctuation), tokenization, stop words removal, lemmatization. Word Embedding used conduct sentiment implementing Word2Vec. After that, One-Against-All (OAA) method with Support Vector Machine (SVM) model make predictions form multi labelling. SVM go through hyperparameter tuning using Randomized Search CV. Then, evaluation carried out Micro Averaged F1 Score assess prediction accuracy Hamming Loss numbers pairs sample label incorrectly classified. Implementation result Word2Vec OAA provide best for data undergoing generalization, lemmatization which stored 100 features model. percentage produced tuned 83.40% 15.13% respectively.
 
 Index Terms— Analysis; Embedding; Word2Vec; One-Against-All; Machine; Challenge; Multi Labelling