Using hidden Markov models to evaluate the quality of discovered process models

Abstract

Hidden Markov Models (HMMs) are a stochastic signal modeling formalism that is actively used in the machine learning community for a wide range of applications such as speech and activity recognition. Efficient techniques exist to learn HMM models from a given data set, and to estimate the data likelihood with respect to a given HMM (i.e., “How probable is it that these data were produced by this HMM?”). The latter enables the evaluation and selection of suitable models. In the domain of process mining the evaluation of models (i.e., “How can we measure the quality of a mined process model?”) is still subject to ongoing research. Because the types of models used in process mining are typically on a higher level of abstraction (i.e., they are more expressive as they, for example, allow to capture concurrency), the problem of model evaluation is challenging. In this paper, we investigate the possibilities and limitations of using HMMs for evaluating process models and compare the resulting quality measurements to the metrics typically used in the process mining domain.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 19 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.