通用多模态Traceformer：整合异构上下文用于流程事件预测

中文摘要 AI 辅助

事件日志广泛产生于各类真实世界流程中，不仅记录了事件活动和时间戳，还包含了多模态的上下文信息。现有的事件序列模型，包括许多时间点过程方法，主要对事件活动和时间戳进行建模，而忽略了异构上下文，例如与单个事件和整个轨迹相关的数值测量、类别属性、文本描述和元数据。在本文中，我们提出了通用多模态Traceformer（UMT），这是一个统一框架，用于将异构流程上下文整合到下一事件预测中。UMT基于Transformer骨干架构，引入了一个通用特征编码器，将多样化的特征类型映射到共享表示空间，并在事件级和轨迹级处理上下文信息。UMT进一步开发了一个逐事件Perceiver模块，该模块动态加权上下文特征，并将其自适应地整合到事件令牌表示中。为了适应到达间隔时间的重尾且可能多模态的分布，UMT在多个时间尺度上表示每个间隔，并联合预测相应的尺度特定数量。在13个真实世界事件日志上的实验表明，UMT在下一事件活动和时间预测方面均优于现有方法。

英文摘要

Event logs arise in a wide range of real-world processes, capturing not only event activities and timestamps but also multi-modal contextual information. Existing event-sequence models, including many temporal point process approaches, primarily model event activities and timestamps while overlooking heterogeneous context, such as numerical measurements, categorical attributes, textual descriptions, and metadata associated with individual events and entire traces. In this paper, we propose Universal Multi-Modal Traceformer (UMT), a unified framework for incorporating heterogeneous process context into next-event prediction. Built on a Transformer backbone, UMT introduces a universal feature encoder that maps diverse feature types into a shared representation space and handles contextual information at both the event and trace levels. UMT further develops a per-event Perceiver module that dynamically weights contextual features and adaptively integrates them into event-token representations. To accommodate the heavy-tailed and potentially multi-modal distribution of inter-arrival times, UMT represents each interval at multiple temporal scales and jointly predicts the corresponding scale-specific quantities. Experiments on 13 real-world event logs show that UMT improves both next-event activity and time prediction over existing approaches.

发表机构

• Celonis

机构由 AI 辅助整理，请以论文原文为准。