@inproceedings{orshansky-etal-2025-hallutree,
    title = "{H}allu{T}ree: Explainable Multi-Hop Hallucination Detection for Abstractive Summarization",
    author = "Orshansky, Daniel  and
      Oomen, Oskar  and
      Agarwal, Naaisha  and
      Lagasse, Ryan",
    editor = "Dong, Yue  and
      Xiao, Wen  and
      Zhang, Haopeng  and
      Zhang, Rui  and
      Ernst, Ori  and
      Wang, Lu  and
      Liu, Fei",
    booktitle = "Proceedings of The 5th New Frontiers in Summarization Workshop",
    month = nov,
    year = "2025",
    address = "Hybrid",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-emnlp/2025.newsum-main.9/",
    pages = "123--134",
    ISBN = "979-8-89176-337-1",
    abstract = "Black-box verifiers for abstractive summaries often struggle with complex claims that require multi-hop reasoning, and they typically provide a single verdict without an interpretable rationale. As a result, it becomes difficult to understand or audit their failures. We address this with HalluTree, a framework that models verification as an interpretable claim tree. HalluTree first decomposes summaries into subclaims, classifying each into two types {--} extractive (directly verifiable against evidence) or inferential (requiring reasoning) {--} which follow distinct verification paths. Extractive claims are robustly verified against evidence using an ensemble of lightweight NLI models. Crucially, inferential claims trigger a process that generates a natural program {--} an explicit reasoning chain that integrates supporting evidence and logical steps {--} which is then executed to determine the claim{'}s validity. Evaluation on the LLM-AggreFact benchmark demonstrates HalluTree{'}s effectiveness: it achieves performance competitive with top-tier black-box models, including Bespoke-MiniCheck, while providing transparent and auditable reasoning programs for every inferential judgment. This combination of competitive accuracy and high interpretability offers a significant advance over opaque, single-classification verifiers. We will publically release code, data, prompts, and other artifacts upon acceptance."
}