Classification and visualization of e-commerce product reviews comparison using support vector machine / Nuwairah Aimi Ahmad Kushairi
E-commerce has experienced significant growth as a platform for online shopping, offering convenience and cost-saving benefits. Especially in Malaysia, Shopee is known to be one of the leading e-commerce platforms, attracting millions of monthly visitors. These days, it has come to our attention tha...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://ir.uitm.edu.my/id/eprint/89008/1/89008.pdf https://ir.uitm.edu.my/id/eprint/89008/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | E-commerce has experienced significant growth as a platform for online shopping, offering convenience and cost-saving benefits. Especially in Malaysia, Shopee is known to be one of the leading e-commerce platforms, attracting millions of monthly visitors. These days, it has come to our attention that online product reviews play a crucial role in influencing consumer behaviour by building trust, identifying customer needs, and improving satisfaction. It was agreed by 97.3% of 186 respondents from a questionnaire survey to rely on product reviews before purchasing any product. Nevertheless, 96.2% of the respondents agreed that not all product reviews are helpful when shopping online. Online customers or shoppers rely on these reviews as decision-making to purchase products. It becomes time-consuming to read through the reviews, especially when it is not product related. Moreover, the large number of reviews could lead to information overload, which exhausts customers to decide. Therefore, this project aims 1) to design a web-based system that can classify the comparison of useful and not useful product reviews from Shopee using the Support Vector Machine (SVM) algorithm and visualize the comparison, 2) to develop the designed system, and 3) to test the functionality and usability of the system. Users can enter a maximum of six product links into the system. The system classifies the reviews into “Useful” or “Not Useful” based on the review text, star rating, duplicated spam, and sentiment score. Afterward, the system recommends the best shop to purchase from and visualizes the reviews to compare products from different shops. The SVM classifier model successfully classified the reviews with an accuracy of 96.8% during the testing stage of the classification. Aside from that, the system was thoroughly evaluated for its functionality, which passed all test cases with expected performance. From the Mann-Whitney U Test in reliability testing, the obtained p-value, 0.008, is below the significance level, p-value<0.05, hence, rejecting the null hypothesis. It means a significant difference exists between the time taken for manual evaluation and the evaluation using the web-based system. As a result, the system has the potential to help users make informed decisions while making purchases. |
|---|