A cost-efficient and explainable model for faithfulness hallucination detection in large language...