A COMPARATIVE STUDY OF MACHINE-LEARNING CLASSIFIERS USING WORD2VEC FOR MYANMAR SOCIAL MEDIA COMMENTS

Abstract

Sentiment Analysis (SA) is also known as opinion mining. The fundamental task of Sentiment Analysis (SA) is to extract, identify and determine the subjective information that is social sentiment in the source text. The application of Sentiment Analysis found in analyzing opinion to classify the class of a document or collection of documents, like blog posts, reviews, news articles and social media feeds like tweets and status updates. This study presents the idea of Sentiment Analysis in term of comparison on three different machine learning techniques, and illustrates how Myanmar Facebook comments are classified by applying word vector representation techniques. In this work, Facebook comments are used for model training and testing. They are collected from Myanmar Celebrity page. These comments are transformed as word vector by using word vector representation techniques – TF-IDF, Word2Vec and Pre-trained Word2Vec. Then, the word vector are trained with three machine learning classifiers for sentiment identification. Three machine learning classifiers are Logistic Regression, Support Vector Machine (SVM) and Random Forest. According to the experimental results of Pre-trained Word2Vec outperformed other two methods of word vector representation technique.