Duration-aware alignment of process traces

Document Type

Conference Proceeding

Publication Date

1-1-2016

Journal

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Volume

9728

DOI

10.1007/978-3-319-41561-1_28

Abstract

© Springer International Publishing Switzerland 2016. Objective: To develop an algorithm for aligning process traces that considers activity duration during alignment and helps derive data-driven insights from workflow data. Methods: We developed a duration-aware trace alignment algorithm as part of a Java application that provides visualization of the alignment. The relative weight of the activity type vs. activity duration during the alignment is an adjustable parameter. We evaluated proportional and logarithmic weights for activity duration. Results: We used duration-aware trace alignment on two real-world medical datasets. Compared with existing context-based alignment algorithm, our results show that duration-aware alignment algorithm achieves higher alignment accuracy and provides more intuitive insights for deviation detection and data visualization. Conclusion: Duration-aware trace alignment improves upon an existing trace alignment approach and offers better alignment accuracy and visualization.

This document is currently not available here.

Share

COinS