AutoLife: Automatic Life Journaling with Smartphones and LLMs

作者：Huatao Xu, Zilin Zeng, Panrong Tong, Mo Li, Mani Srivastava · 发表于：Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies · 年份：2025 · DOI：10.1145/3770683 · 被引用次数：3 · 研究领域：Context-Aware Activity Recognition Systems、Personal Information Management and User Behavior、Innovative Human-Technology Interaction

This paper introduces a novel mobile sensing application - life journaling - designed to generate semantic descriptions of users' daily lives. We present AutoLife, an automatic life journaling system based on commercial smartphones. AutoLife only inputs low-cost sensor data (without photos or audio) from smartphones and can automatically generate comprehensive life journals for users. To achieve this, we first derive time, motion, and location contexts from multimodal sensor data, and harness the zero-shot capabilities of Large Language Models (LLMs), enriched with commonsense knowledge about human lives, to interpret diverse contexts and generate life journals. To manage the task complexity and long sensing duration, a multilayer framework is proposed, which decomposes tasks and seamlessly integrates LLMs with other techniques for life journaling. This study establishes a real-life dataset as a benchmark and extensive experiment results demonstrate that AutoLife produces accurate and reliable life journals. We also demonstrate the real-world application of AutoLife by integrating it into a digital journaling app that automatically generates journals for users.