SimTrace generates synthetic user clickstreams for better online modeling
SimTrace: Grounded Multimodal User Trajectories Generation for Online User Modeling
Artificial Intelligence
Summary
Many companies need detailed records of how users interact with websites to improve services, but real user data is hard to get due to privacy and limited traffic. The authors created SimTrace, a system that makes realistic fake user activity based on real usage while protecting privacy. This synthetic data helps build models that predict user actions and improve recommendations almost as well as real data. Adding synthetic data to real data can even make predictions more accurate.
What this means in practice
- •For e-commerce developers: Create realistic user interaction data to improve product recommendations and purchase predictions without compromising privacy.
- •For user interface designers: Simulate detailed user activity in web environments to test and evaluate interface changes before deployment.
Authors
Yunan Lu, Shuang Xie, Meghna Allamudi, Mingyu Zhao, Han Li, Lingyun Wang, Zhou Yu
Abstract
Virtual clients offer a cost-effective approach to support applications such as A/B testing, recommender system development, and interface evaluation. However, building them requires access to large-scale, semantically faithful, fine-grained online user trajectories. These data are difficult to obtain because proprietary logs are subject to privacy restrictions and small businesses often lack sufficient traffic. Consequently, existing public datasets either abstract away fine-grained user interaction details or preserve rich context but remain platform-specific and small-scale. To address this gap, we propose SimTrace, a framework that generates faithful, fine-grained synthetic multimodal clickstreams through a computer-use client agent that is grounded in real user trajectories and the given web environment. SimTrace anonymizes real interactions and constructs a simulated twin of the given web environment, then uses both to generate synthetic interaction trajectories. Each action is paired with its corresponding web observations and user context, yielding a shareable alternative to confidential logs for developing computer-use agent-style virtual clients. We apply SimTrace to an e-commerce setting and evaluate both its fidelity and downstream utility. SimTrace outperforms competing baselines on 7 out of 8 fidelity metrics. Models trained on synthetic data achieve performance comparable to those trained on real data on downstream tasks such as purchase prediction and recommendation. For next action prediction task, augmenting real data with synthetic data further improves accuracy by 11.0% relative to training on real data alone. We release SimTrace as an open-source package to facilitate research on online user behavior modeling.