Public Sentiment Towards Mandatory Halal Certification: A Large Language Model (LLM) Approach

Abstract

This study analyzes public sentiment towards mandatory halal certification in Indonesia, as mandated by Law No. 33/2014 and its revision in Government Regulation No. 39/2021. Using the Large Language Model (LLM) approach, sentiment analysis was conducted on a dataset consisting of 320 samples of headlines from various electronic media platforms, published between 2019 and 2023. The LLM model, employing the RoBERTa architecture, was trained on an Indonesian language dataset and optimized for sentiment classification tasks. Data preprocessing included web scraping, data cleansing, and text vectorization using Term Frequency-Inverse Document Frequency (TF-IDF) techniques and cosine-similarity. The model demonstrated a confidence score of the classifications mean of 87.35% and median 96.12% in classifying the news headlines. Results revealed a predominant positive sentiment (57%) towards halal certification, indicating public awareness of its significances. However, negative sentiments (26%) highlighted challenges faced by Small and Medium Enterprises (SMEs), including high costs and lack of understanding about the certification process. The temporal analysis showed an increase in halal-related news coverage following the implementation of new regulations. This study contributes to the understanding of public perception towards regulatory changes in the halal industry and demonstrates the effectiveness of LLM-based sentiment analysis in comprehending public opinions. The findings provide valuable insights for policymakers and businesses in addressing the potential and challenges in implementing mandatory halal certification.