发表机构
Yale University; IC3; Staples High School（耶鲁大学; IC3; 斯台普斯高中）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对机器人检测中标签匮乏问题，提出FISSION方法，通过拆分账户活动生成标签以学习嵌入，在维基百科和Twitter/X数据集上优于现有方法。
AI 中文摘要
机器人账户和协调性影响力操作通常通过启发式方法被发现，这导致用于训练检测系统的可靠地面真值标签匮乏。为了解决这一挑战，我们研究了一个自然问题：我们能否生成标签来辅助学习嵌入，使得来自同一协调操作的机器人和账户在嵌入空间中彼此接近？我们提出了FISSION，一种通过将每个账户的活动拆分为正标签子账户来生成标签的方法。基于这一标签来源，我们训练检测模型，以保留在正子账户中反复出现的行为规律性。我们评估了FISSION，并展示其在检测维基百科的傀儡账户和Twitter/X机器人方面优于先前方法。
英文摘要
Bot accounts and coordinated influence operations are often discovered via heuristic methods, leaving a dearth of reliable ground-truth labels for training detection systems. To address this challenge, we study a natural question: can we generate labels to assist in learning embeddings in which bots and accounts from the same coordinated operation are close? We present FISSION, a method to generate labels by splitting each account's activity into positively labeled sub-accounts. Given this label source, we train detection models which preserve behavioral regularities recurring across positive sub-accounts. We evaluate FISSION and show it outperforms prior methods in detecting Wikipedia sockpuppets and Twitter/X bots.
Comments37 pages, 10 figures, 20 tables
