发表机构
University of Florida（佛罗里达大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出一种单次传递的联合架构用于目标立场抽取，减少近50%参数且仅牺牲4-7个F1点，并证明保留目标提及可提升至少6个F1点，便于下游应用集成。
AI 中文摘要
目标立场抽取（Target-Stance Extraction, TSE）是一项预测作者写作的目标（或主题）以及作者对该目标所持立场的任务。现有的TSE方法采用两个独立神经模型的顺序流水线：一个用于识别目标，另一个用于确定立场。我们提出了一种单次传递的联合架构，在前向传播中同时预测两者，将可训练参数减少了近50%，而性能仅牺牲4-7个F1点。我们进一步证明，标准的目标清洗做法会人为地抑制目标预测的准确性。保留明确的目标提及（如实际部署中的情况）在目标分类和目标生成两种设置下均能将F1分数至少提升6个点。这些改进使得TSE在下游应用（如公众舆论追踪）中的集成显著更加容易。
英文摘要
Target-Stance Extraction (TSE) is the task of predicting both the target (or topic) of an author's writing and the author's stance toward it. Existing approaches to TSE use a sequential pipeline of two separate neural models: one to identify the target and another to determine the stance. We present a one-pass, joint architecture that predicts both in a single forward pass, reducing trainable parameters by nearly 50% with only a 4-7 F1 point tradeoff in performance. We further demonstrate that standard target-scrubbing practices artificially suppress target prediction accuracy. Retaining explicit target mentions, as in real-world deployments, improves F1 by at least 6 points across both target classification and target generation settings. These improvements allow for significantly easier integration of TSE in downstream applications such as public opinion tracking.
Comments10 pages, 3 figures, 7 tables
