Abstract
The growth of TikTok has reshaped social media marketing, with influencer-driven content playing a crucial role in consumer engagement and purchasing decisions. In the Malaysian Beauty and Personal Care category, TikTok comments serve as direct consumer feedback, yet extracting meaningful insights from this highly expressive data remains challenging. Traditional sentiment analysis methods struggle with multilingual text, slang, abbreviations, and emojis, limiting their effectiveness in interpreting user sentiment. This study addresses these challenges by leveraging GPT-based sentiment analysis to analyze TikTok comments, examining sentiment trends, linguistic patterns, and their correlation with influencer revenue. The study focuses on the top 20 highest-revenue influencers within the Malaysian Beauty and Personal Care category, collecting 34,597 comments from 3,912 videos using Apify’s TikTok scraping API. The dataset was preprocessed using GPT-based text normalization, slang resolution, and emoji-to-text conversion, ensuring consistency in sentiment classification. It categorized comments into Positive, Neutral, or Negative, followed by a detailed examination of frequently used words and sentiment patterns across different influencers.
Metadata
Item Type: | Article |
---|---|
Creators: | Creators Email / ID Num. Ahmad Asmawi, Muhammad Akmal Hakim UNSPECIFIED Isawasan, Pradeep pradeep@uitm.edu.my Shamugam, Lalitha UNSPECIFIED Ahmad Salleh, Khairulliza UNSPECIFIED Savita, K.S. UNSPECIFIED |
Subjects: | Q Science > QA Mathematics > Analysis > Analytical methods used in the solution of physical problems |
Divisions: | Universiti Teknologi MARA, Perak > Tapah Campus > Faculty of Computer and Mathematical Sciences |
Journal or Publication Title: | e-Academia Journal |
UiTM Journal Collections: | UiTM Journals > e-Academia Journal (e-AJ) |
ISSN: | 2289 - 6589 |
Volume: | 14 |
Number: | 1 |
Page Range: | pp. 57-72 |
Keywords: | Business intelligence, Social media analytics, TikTok, Sentiment analysis, User generated content |
Date: | June 2025 |
URI: | https://ir.uitm.edu.my/id/eprint/118867 |