Analyzing Public Perceptions of MSME Digitalization Through TikTok Comments Using IndoBERT

An NLP-Based Sentiment Analysis Approach for Indonesian Social Media Data

Abstract

The digitalization of MSMEs is one of the important efforts to increase business competitiveness and sustainability in the digital economy era. Various information related to the digitization of MSMEs circulating on social media has generated various responses from the public, especially on the TikTok platform. This study aims to analyze the sentiment of TikTok users' comments on the digitization of MSMEs using the IndoBERT model. The research data was obtained through the TikTok comment scraping process using Apify in the period from December 1, 2024, to May 24, 2026. After going through the manual labeling and pre-processing stages of data, a total of 1,042 comments were obtained, classified into three sentiment categories, namely positive, negative, and neutral. The research stages include preprocessing, tokenization, label mapping, data splitting using the stratified splitting method, and fine-tuning the IndoBERT model for three epochs. The results of the evaluation showed that the model achieved an accuracy value of 95.19%, a precision of 95.36%, a recall of 95.19%, and an F1-score of 95.25%. The results of the sentiment analysis showed that neutral sentiment dominated with a percentage of 82.73%, followed by negative sentiment of 9.21% and positive sentiment of 8.06%. These findings indicate that most users are still at the stage of seeking information, asking questions, or sharing experiences related to the digitization of MSMEs. This study proves that IndoBERT has a very good performance in classifying Indonesian sentiment on social media data.

Keywords: Sentiment Analysis, Digitization of MSMEs, TikTok, IndoBERT, Natural Language Processing.