Microflows: Leveraging Process Mining and an Automated Constraint Recommender for Microflow Modeling

@inproceedings{Oberhauser2018,
    Author = "Oberhauser, Roy and Stigler, Sebastian",
    Editor = "Shishkov, Boris",
    Title = "{Microflows: Leveraging Process Mining and an Automated Constraint Recommender for Microflow Modeling}",
    Booktitle = "Business Modeling and Software Design",
    Year = "2018",
    Address = "Cham",
    Pages = "25--48",
    Publisher = "Springer International Publishing",
    Abstract = "Businesses and software development processes alike are being challenged by the digital transformation and agility trend. Business processes are increasingly being automated yet are also expected to be agile. Current business process modeling is typically labor-intensive and results in rigid process models. For larger processes it becomes arduous to consider all possible process variations and enactment circumstances. Contemporaneously, in software development microservices have become a popular software architectural style for partitioning business logic into fine-grained services accessible via lightweight protocols which can be rapidly and individually developed by small teams and flexibly (re)deployed. This results in an increasing number of available services and a much more dynamic IT service landscape. Thus, a more dynamic form of modeling, integration, and orchestration of these microservices with business processes is needed. This paper describes agile business process modeling with Microflows, an automatic lightweight declarative approach for the workflow-centric orchestration of semantically-annotated microservices using agent-based clients, graph-based methods, and the lightweight semantic vocabularies JSON-LD and Hydra. A graphical modeling tool supports Microflow modeling and provides dynamic constraint and microservice recommendations via a recommender service using machine learning of domain-categorized Microflows. To be able to utilize existing process model knowledge, a case study shows how Microflow constraints can be automatically extracted from existing Business Process Modeling Notation (BPMN) process files and transformed into flexible Microflow constraints, which can then be used to train the recommendation service. Further, it describes process mining of Microflow execution logs to automatically extract BPMN models and automated recovery for errors occurring during enactment.",
    ISBN = "978-3-319-78428-1"
}