Hallucination Detection and Mitigation in Scientific Text Simplification using Ensemble Approaches: DS@GT at CLEF 2025 SimpleText
This addresses the challenge of ensuring accuracy in simplified scientific texts for researchers and educators, though it appears incremental as it builds on existing ensemble and LLM methods.
The paper tackles the problem of detecting and mitigating hallucinations in scientific text simplification by developing an ensemble framework that combines multiple strategies including BERT-based classifiers and LLM reasoning, achieving enhanced robustness in distortion detection.
In this paper, we describe our methodology for the CLEF 2025 SimpleText Task 2, which focuses on detecting and evaluating creative generation and information distortion in scientific text simplification. Our solution integrates multiple strategies: we construct an ensemble framework that leverages BERT-based classifier, semantic similarity measure, natural language inference model, and large language model (LLM) reasoning. These diverse signals are combined using meta-classifiers to enhance the robustness of spurious and distortion detection. Additionally, for grounded generation, we employ an LLM-based post-editing system that revises simplifications based on the original input texts.