Textdokument
Addressing the Log Representativeness Problem using Species Discovery (Extended Abstract)
Vorschaubild nicht verfügbar
Volltext URI
Dokumententyp
Dateien
Zusatzinformation
Datum
2024
Autor:innen
Zeitschriftentitel
ISSN der Zeitschrift
Bandtitel
Quelle
Verlag
Gesellschaft für Informatik, Bonn
Zusammenfassung
Event logs are generated during the enactment of process-centric information systems and form the basis for optimization, monitoring, and enhancement initiatives of said systems. As such, they enable a data-driven and unbiased evaluation of the as-is state of the underlying processes. Yet, since at any time, event logs represent merely a sample of the whole possible behaviour of the information system, insights are only actionable should the event log be representative of the information system from which it is derived. Therefore, the question arises of how the representativeness of an event log$L with respect to its generative system P can be quantified, given that only L is present. In this work, we argue, that representativeness of an event log needs to be assessed with an intended analysis question in mind and discuss log completeness as one important facet of representativeness. We show how established estimators from biodiversity research can be utilized to quantify log completeness.