@inproceedings{tran-etal-2026-llm,
    title = "{LLM}-{INSTRUCT} at {UZH} Shared Task 2026: Constraint-Aware Retrieval and Selective Debate for Paragraph-Level Argument Mining",
    author = "Tran, Phuong Huu Vu  and
      Vo, Long Minh  and
      Le, Son Nguyen Minh  and
      Van, Hoang",
    editor = "Elaraby, Mohamed  and
      Hautli-Janisz, Annette  and
      Romberg, Julia  and
      Musi, Elena  and
      Ruggeri, Federico  and
      Lawrence, John",
    booktitle = "Proceedings of the 13th Workshop on Argument Mining and Reasoning",
    month = jul,
    year = "2026",
    address = "San Diego, California, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/bulk-corrections-2026-07-02/2026.argmining-1.11/",
    doi = "10.18653/v1/2026.argmining-1.11",
    pages = "99--104",
    ISBN = "979-8-89176-399-9",
    abstract = "We present LLM-INSTRUCT, the winning system for the UZH Shared Task at ArgMining 2026 on paragraph-level argument mining in UN and UNESCO resolutions. The task requires paragraph-type classification, prediction of a subset of 141 official tags, and directed relation prediction under a strict JSON schema setting using only open-weight models up to 8B parameters. We frame the task as constrained structured prediction. The system first narrows the candidate tag space with metadata-aware dense retrieval, then applies constrained decoding with per-dimension caps, and escalates only uncertain cases to a three-agent debate branch."
}