Muhammad Farhan Firdaus, Hairol Zaman (2023) Sentiment Analysis For E-Commerce Clothing Reviews Using Bidirectional Recurrent Neural Network. Faculty of Computing, Universiti Malaysia Pahang Al-Sultan Abdullah.
|
Pdf
CB20153.pdf - Accepted Version Download (3MB) | Preview |
Abstract
Today's marketing methods place a high priority on comprehending client emotions. Companies will gain insight into how customers view their goods and/or services, and they will get ideas on how to enhance their offerings. Traditional methods of sales and business are not as effective as the e-commerce approach. It is such a hassle for customers to walk into the retail store and purchase their product needs. It is a waste of time and energy for today which world is full of technology. This study makes an effort to comprehend the relationship between several factors in customer reviews on an online store selling women's clothes. It also aims to categorize each review according to whether it recommends the product under consideration and whether it expresses a positive, negative, or neutral attitude. Thus, this study proposed a bidirectional recurrent neural network (RNN) with a long-short-term memory unit (LSTM) for sentiment classification. Results have indicated that a major predictor of a high sentiment score is a recommendation, and vice versa. Ratings in product reviews, on the other hand, are hazy predictors of sentiment scores. Additionally, we discovered that the bidirectional LSTM achieved an F1-score of 0.93 for sentiment classification.
| Item Type: | Undergraduates Project Papers |
|---|---|
| Additional Information: | SV: Ts. Dr. Zuriani Mustaffa |
| Uncontrolled Keywords: | e-commerce, bidirectional recurrent neural network (RNN), long-short-term memory unit (LSTM) |
| Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science |
| Faculty/Division: | Faculty of Computing |
| Depositing User: | Mr. Nik Ahmad Nasyrun Nik Abd Malik |
| Date Deposited: | 03 Apr 2024 07:43 |
| Last Modified: | 03 Apr 2024 07:43 |
| URI: | http://umpir.ump.edu.my/id/eprint/40875 |
| Download Statistic: | View Download Statistics |
Actions (login required)
 |
View Item |