Comparing Machine Learning and Deep Learning Based Approaches to Detect Customer Sentiment from Product Reviews

Citation

Lim, Sin Li and Foo, Lee Kien and Chua, Sook Ling (2023) Comparing Machine Learning and Deep Learning Based Approaches to Detect Customer Sentiment from Product Reviews. Journal of System and Management Sciences, 13 (2). pp. 101-110. ISSN 1816-6075, 1818-0523

[img] Text
Vol.pdf - Published Version
Restricted to Repository staff only

Download (338kB)

Abstract

Product review is way for customers to express their sentiments towards a product. Sentiment analysis can be performed to gain insights from product reviews and help to improve the sale-ability of a product. This research aims to perform sentiment analysis on reviews of electronic products. In our study, we compared two methods: one is based on deep learning and another is based on machine learning. The public data of product reviews obtained from Amazon and BestBuy were used in this study. We compared a deep learning method and a machine learning method by first investigated the impact of training data selected using different sampling techniques. Then we examined the effect of hyper-parameter tuning on these learning algorithms. The resulted best models were then compared to the baseline model for sentiment analysis. The experimental results showed that our models performed better than the baseline model in terms of accuracy and F1. The experimental outcome suggests that customer’s sentiment towards a product can be captured by applying deep learning and machine learning on product reviews.

Item Type: Article
Uncontrolled Keywords: Sentiment analysis, convolutional neural network, sampling techniques, support vector machine.
Subjects: Q Science > QA Mathematics > QA71-90 Instruments and machines > QA75-76.95 Calculating machines
Divisions: Faculty of Computing and Informatics (FCI)
Depositing User: Ms Nurul Iqtiani Ahmad
Date Deposited: 04 Jul 2023 03:26
Last Modified: 04 Jul 2023 03:26
URII: http://shdl.mmu.edu.my/id/eprint/11525

Downloads

Downloads per month over past year

View ItemEdit (login required)