Presentation Information

4:00 PM - 4:15 PM JST(7:00 AM - 7:15 AM UTC)Candidate for Academic Incentive Award

[B-14-22]A Method for Generating Evidence-Based Task and Case Segmentation Candidates from GUI Operation Logs and Business Documents

◎Misa Fukai¹, Sayaka Yagi¹, Toshiki Kadoya¹, Shinji Ogawa¹, Masaki Ishiyama¹ (1. NTT, Inc)

Keywords:

Operation Logs,Task and Case Segmentation,Business Documents

This paper proposes a method for generating case and task segmentation candidates from GUI operation logs by referring to business documents. To promote business automation and efficiency improvement, it is important to organize actual work practices into meaningful business units. However, in real business operations, task and case boundaries are often ambiguous due to interruptions and parallel work, and an appropriate segmentation granularity may vary depending on the analysis purpose. The proposed method preprocesses GUI operation logs and business manuals, estimates case anchors based on identifiers such as application or ticket IDs, and estimates task boundary candidates based on application switching, URL changes, idle time, and related cues. It then combines these estimates to generate multiple segmentation candidates and associates each candidate with evidence such as corresponding log intervals and document passages. This enables users to compare and verify candidates and select segmentation results suitable for business analysis.

Back to Session information