Mining for workarounds in text fields with clustering algorithms

  • J. (Janneke) Spronk

Student thesis: Master's Thesis

Abstract

This study describes how text mining and clustering algorithms can detect workarounds in a database. Finding workarounds can reduce risk and improve IT system resilience by guiding subsequent system improvement. Workarounds are methods used by employees that are not intended by the IT department. Analyzing workarounds provides important information about opportunities and threats in business processes and IT systems. Organizations base their choices of improvement on the intended use of IT, and do not include workarounds in their argumentation for change. With knowledge of workarounds, better choices on IT improvements can be made and organization resilience is improved. Therefore, a Resilience Mining Thesis Circle started to explore the possibilities of using algorithms to find workarounds in a database. This study is part of this Thesis Circle, and focusses on clustering algorithms to detect workarounds in text fields in IT systems. The research is conducted on data of the department of Collective Pensions of Achmea, a large insurance company in the Netherlands. We find that k-Means (kernel) and Agglomerative clustering are useful algorithms that can detect workarounds in text fields. Combined with the findings of the other members of the Resilience Mining Thesis Circle, we provide an overview of the possibilities for using algorithms to detect workarounds in IT systems.
Date of Award25 Jan 2020
Original languageEnglish
SupervisorLloyd Rutledge (Examinator) & Guy Janssens (Co-assessor)

Keywords

  • Workarounds
  • database
  • IT systems
  • algorithm
  • clustering
  • text mining
  • k-Means
  • DBSCAN
  • k-Medoids
  • Agglomerative clustering

Cite this

'