Title: Hybrid Process Mining: Combining Imperative & Declarative Paradigms
Speaker: Christoffer Olling Back (PhD Fellow, Department of Computer Science, University of Copenhagen)
Modern organisations generate enormous amounts of data related to their business processes via IT systems such as Enterprise Resource Planning (ERP) systems. Mining process related data presents the opportunity for a truly data-driven approach to streamlining operations and building an intelligent enterprise. Depending on the use case, process mining may be employed with the aim of detecting deviations from a well-defined process, identifying areas for process enhancement (detecting bottlenecks, unnecessary rework), and in the case of pure process discovery – where the underlying process is not known – the task essentially becomes comparable to unsupervised learning.
In this talk, I will begin with a brief introduction to process mining generally, positioning the field in relation to data mining and machine learning. Afterwards, I will discuss different process modelling paradigms in more detail, and their strengths and weaknesses. In particular, the distinction between declarative (constraint-based) and imperative (flow-based) languages, and how they can be combined to form hybrid process models which combine the strengths of both. Time permitting, I will discuss my ongoing work on developing process mining algorithms which generate hybrid models by identifying pockets of regularity using an information theoretic approach, as well as challenges in empirically evaluating process mining algorithms.