Document Type

Dissertation

Rights

This item is available under a Creative Commons License for non-commercial use only

Disciplines

Computer Sciences

Publication Details

A dissertation submitted in partial fulfilment of the requirements of Technological University Dublin for the degree of M.Sc. in Computer Science (Data Science) 2021

Abstract

Sentiment Analysis also known as opinion mining is a type of text research that analyses people’s opinions expressed in written language. Sentiment analysis brings together various research areas such as Natural Language Processing (NLP), Data Mining, and Text Mining, and is fast becoming of major importance to companies and organizations as it is started to incorporate online commerce data for analysis. Often the data on which sentiment analysis is performed will be reviews. The data can range from reviews of a small product to a big multinational corporation. The goal of performing sentiment analysis is to extract information from those reviews to gauge public opinion for market research, monitor brand and product reputation, and understand customer experiences. Reviews written on the online platform are often in the form of free text and they do not have any standard structure. Dealing with unstructured data is a challenging problem. Sentiment analysis can be done at different levels, and the focus of this research is on aspect-level sentiment analysis. In aspect-level sentiment analysis, there are two tasks that need to be addressed. The first task is aspect identification which is the process of discovering those attributes of the object that people are commenting on. These attributes of the object are called aspects. The second task is the sentiment classification of those reviews using these extracted aspects. For the sentiment analysis, transformer-based pre-trained models such as BERT (Bidirectional Encoder Representations from Transformers) and RoBERTa (A robustly optimized BERT) are used in this research as they make use of embedding vector space that is rich in context. The purpose of this research is to propose a framework for extracting the aspects from the data which can be applied to these pre-trained models. For the first part of the experiment, both the BERT and RoBERTa models are developed without the aspect-based approach. For the second part of the experiment, the aspect-based approach is applied to the same models and their results are compared and evaluated against the equivalent models. The experiment results show that aspect-based approach has increased the performance of the models by almost 1% than the traditional models and the BERT model with the aspect-based approach had the highest accuracy and performance among all the models evaluated in this research..

DOI

https://doi.org/10.21427/3w9n-we77


Share

COinS