A Logic Framework for Incremental Learning of Process Models

Abstract

Standardized processes are important for correctly carrying out activities in an organization. Often the procedures they describe are already in operation, and the need is to understand and formalize them in a model that can support their analysis, replication and enforcement. Manually building these models is complex, costly and error-prone. Hence, the interest in automatically learning them from examples of actual procedures. Desirable options are incrementality in learning and adapting the models, and the ability to express triggers and conditions on the tasks that make up the workflow. This paper proposes a framework based on First-Order Logic that solves many shortcomings of previous approaches to this problem in the literature, allowing to deal with complex domains in a powerful and flexible way. Indeed, First-Order Logic provides a single, comprehensive and expressive representation and manipulation environment for supporting all of the above requirements. A purposely devised experimental evaluation confirms the effectiveness and efficiency of the proposed solution.processes and to enact them. The availability of a suitable process model can determine production and economic success of a company. While most information for properly setting up these models comes from the practitioners that day by day carry out and improve the procedures, formal models of activities are typically produced by supervisors and managers, a gap that often causes models not to perfectly fit the practice. Additionally, producing the models is in itself a complex, costly and error-prone activity [22]. A related problem is the need of adapting and fine-tuning existing models, when available, in dynamic environments (approached in [14] by allowing modifications to the workflow model). Hence, the strong motivation to study processes and techniques for modeling them, and the need of tools that can (at least partially) automatically learn them. For instance, companies that do not have a process management yet can obtain this way an initial model from examples of current practice, to be later refined and finetuned by experts; on the other hand, companies that have developed an abstract model can compare it to the current practice, or exploit it to supervise subsequent process executions or to generate simulated executions and check their correctness or compliance to their policy.In the following, we present a novel approach to learning process models in a perspective of strict comparison to previous proposals, in order to show both their similarities and their differences. The rest of the paper is organized as follows. Section 2 introduces basic notions about the application field, and Section 3 describes previous approaches existing in the literature to tackle these problems. Then, the proposed solution (representational framework and inference techniques) is described in Section 4, also pointing out noteworthy features with respect to previous proposals. Section 5 discusses possible exploitations of the learned models, and Section 6 re...

Activity prediction in process mining using the WoMan framework

“…The same holds for GPItaly, except for concurrency. The chess datasets are characterized by very high concurrency: each game starts with 32 concurrent activities, a number which is beyond the reach of many current process mining systems (Ferilli and Esposito 2013). This number progressively decreases (but remains still high) as long as the game proceeds.…”

“…The WoMan framework (Ferilli and Esposito 2013;Ferilli 2014) lies at the intersection between Declarative Process Mining and Inductive Logic Programming (ILP) (Muggleton 1991). Indeed, it pervasively uses First-Order Logic as a representation formalism, that provides a great expressiveness potential and allows one to describe contextual information using relationships.…”

“…Given these motivations, we focused on the activity prediction task and developed two approaches to carry it out. Both are based on the WoMan framework for workflow management (Ferilli and Esposito 2013;Ferilli 2014), which proved able to outperform other state-of-the-art systems when dealing with very complex processes. The former (the heuristic approach) is the one originally adopted by WoMan (Ferilli et al 2016).…”

Partial-order-based process mining: a survey and outlook

“…In [52], an event log is translated to a first-order logic expression, which is subsequently used to update a workflow model incrementally. In [53,54], the authors propose to learn labeled partial orders which are subsequently converted into event structures.…”