FAKTA: An Indonesian Hoax Detection System Based on a Hybrid of LSTM, Large Language Models, and Evidence Retrieval

Published:

Final-year BINUS student capstone, supervised by Gabriel Asael Tarigan. FAKTA models hoax detection along two orthogonal axes - deceptive writing style (bidirectional LSTM plus linguistic features) and correspondence with credible sources (atomic claim decomposition, hybrid BM25 and multilingual dense retrieval, and strict evidence adjudication).