LLM-INSTRUCT at UZH Shared Task 2026: Constraint-Aware Retrieval and Selective Debate for Paragraph-Level Argument Mining
LLM-INSTRUCT at UZH Shared Task 2026: Constraint-Aware Retrieval and Selective Debate for Paragraph-Level Argument Mining
LLM-INSTRUCT 在 UZH 2026 共享任务中的表现：用于段落级论证挖掘的约束感知检索与选择性辩论方法
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, escalates only uncertain cases to a three-agent debate branch, and finally validates the output schema.
摘要： 我们介绍了 LLM-INSTRUCT 系统，该系统在 ArgMining 2026 的 UZH 共享任务中获得冠军，专注于联合国和联合国教科文组织决议中的段落级论证挖掘。该任务要求在严格的 JSON 模式设置下，仅使用参数量不超过 8B 的开源权重模型，完成段落类型分类、141 个官方标签子集的预测以及有向关系预测。我们将此任务定义为约束结构化预测问题。该系统首先通过元数据感知的密集检索缩小候选标签空间，随后应用带有维度上限的约束解码，仅将不确定的案例升级至三智能体辩论分支进行处理，最后对输出模式进行验证。
On the official leaderboard, LLM-INSTRUCT ranked 1st overall, with 1st in F1 and 5th in LLM-as-a-Judge. During development, our configuration search further improved Task 1b Micro-F1 from 35.83% to 40.08% while keeping the internal Task 2 score at 4.421. The main lesson is simple: reducing the decision space before generation improves both accuracy and submission robustness. Our code and supporting scripts are publicly available at: this https URL.
在官方排行榜上，LLM-INSTRUCT 综合排名第一，其中 F1 分数位列第一，LLM-as-a-Judge（大模型作为评判者）排名第五。在开发过程中，我们的配置搜索将任务 1b 的 Micro-F1 分数从 35.83% 提升至 40.08%，同时保持内部任务 2 的得分为 4.421。核心经验很简单：在生成前缩小决策空间可以同时提高准确性和提交结果的稳健性。我们的代码和配套脚本已公开，地址为：[此链接]。