Abstract
Analysis of user interaction logs can be challenging because manual review may not be scalable and aggregated data analysis can obscure the chronological context of user behavior. Systems and methods for automated analysis can use a structured agentic workflow. This workflow can programmatically extract raw, time-ordered session logs and can utilize a large language model, guided by a prompt engineering subsystem, to perform analytical tasks such as classification, summarization, and reasoning on the sequential data. This approach can facilitate scalable and standardized analysis of user behavior by preserving the temporal detail of individual sessions, which may enable AI agent for applications such as metric debugging, user experience issue triage, and longitudinal journey synthesis, including the analysis of multimodal data such as user interface screenshots.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Li, Sha and Yu, Chenjie, "Automated Analysis of User Logs Using Large Language Models in an Agentic Workflow", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/9199