Using Multilevel Business Artifacts for Knowledge Management in Analytics Projects

Autoren
S. Staudinger, C. Schütz, M. Schrefl
Paper
Stau23a (2023)
Zitat
Proceedings of the 10th International Workshop on Multi-Level Modeling (MULTI 2023) in conjunction with the 26th International Conference on Model Driven Engineering Languages and Systems (MODELS 2023), October 1-6, 2023, Västeras, Sweden, Website MULTI: https://jku-win-dke.github.io/MULTI2023/, IEEE Press, pp. 689-698 (10 pages), DOI:10.1109/MODELS-C59198.2023.00110, 2023.
Ressourcen
Kopie  (Senden Sie ein Email mit  Stau23a  als Betreff an dke.win@jku.at um diese Kopie zu erhalten)

Kurzfassung (Englisch)

Analytics projects often follow a generic process model, which maps out the main stages and tasks for conducting an analytics project while granting leeway to the project manager regarding the specific execution. A generic process model is instantiated by various organizations for projects applying different types of analytics - descriptive, predictive, prescriptive, etc. - on different use cases in various domains, using vastly different data. Each organization, each type of analytics, and each individual project thus requires a customized process tailored to the specific needs of the organization, type of analytics, and individual project. At each stage of a data analytics project, the project team has to assess the use case (analytics problem) and determine the course of action. Proper documentation of assessment and course of action, i.e., the design decisions and the underlying motivations, facilitates development in the subsequent stages and tasks as well as after deployment when using the developed system. In this paper, we present a use case for multilevel modeling, namely the documentation of knowledge related to analytics projects and data analyses, which are processes aimed at finding patterns in data. We employ the concept of multilevel business artifact, which allows for the representation of data and life cycle models in a single object at multiple levels of abstraction while granting the flexibility to specialize models in objects at lower levels. We use the real-world problem of flight delay prediction as a running example to illustrate the use of multilevel business artifacts for knowledge management in analytics projects.