Habib, Irham (2025) Evaluasi Kinerja Indober untuk Analisis Sentimen Teks Bahasa Minangkabau. Other thesis, Universitas Islam Riau.

Text skripsi_203510472_watermark.pdf - Submitted Version Restricted to Registered users only Download (7MB) | Request a copy

Abstract

This study aims to evaluate the performance of the IndoBERT model in performing sentiment analysis on Minangkabau-language texts. The Minangkabau language is considered a low-resource language, presenting unique challenges in Natural Language Processing (NLP). This research applies a Fine-Tuning approach to the IndoBERT model—originally trained on Indonesian corpora—so that it can adapt to the distinctive structure and vocabulary of the Minangkabau language. The dataset used consists of parallel Minangkabau and Indonesian sentence pairs, with sentiment labeling conducted through zero-shot prompting using ChatGPT. The model's performance is evaluated using accuracy, Precision, Recall, and F1-score metrics. The results show that the model achieves an accuracy of 83% and an F1score of 82%, demonstrating good performance in classifying both positive and negative sentiments. This study indicates that Fine-Tuning transformer-based models like IndoBERT holds great potential for advancing NLP technologies for local languages in Indonesia.

Actions (login required)

View Item