Open AccessOnline concept drift detection, localization andcharacterization using trace clustering
Author(s) -
Rafael Gaspar de Sousa,
Sarajane Marques Peres
Publication year - 2020
Language(s) - English
Resource type - Conference proceedings
DOI - 10.5753/sbsi.2020.13122
Subject(s) - cluster analysis , trace (psycholinguistics) , concept drift , computer science , flexibility (engineering) , adaptability , data mining , process (computing) , process mining , biclustering , artificial intelligence , business process , machine learning , correlation clustering , data stream mining , work in process , canopy clustering algorithm , mathematics , engineering , business process management , statistics , philosophy , operations management , ecology , linguistics , biology , operating system
Most process mining techniques assume stationary processes and are not well equipped to deal with concept drift. Online detection, localization and characterization of concept drift in business processes can support process mining techniques and analysts to improve organizations flexibility and adaptability. In this research, we propose a method to detect, locate and characterize concept drift in an online setting using trace clustering. The hypothesis is that the method can benefit from the trace clustering capacity to simplify complex problems through grouping similar patterns. In preliminary experiments, trace clustering was performed in a windowing setting showing that concept drift can be detected by analyzing the variation of clustering over time.