Search

Automatically discovering a process model from an event log is the prime problem in process mining. This task is so far approached as an unsupervised learning problem through graph synthesis algorithms. Algorithmic design decisions and heuristics allow for efficiently finding models in a reduced search space. However, design decisions and heuristics are derived from assumptions about how a given behavioral description—an event log—translates into a process model and were not learned from actual models which introduce biases in the solutions. In this paper, we explore the problem of supervised learning of a process discovery technique. We introduce a technique for training an ML-based model using graph convolutional neural networks, which translates a given input event log into a sound Petri net. We show that training this model on synthetically generated pairs of input logs and output models allows it to translate previously unseen synthetic and several real-life event logs into sound, arbitrarily structured models of comparable accuracy and simplicity as existing state of the art techniques in imperative mining. We analyze the limitations of the proposed technique and outline alleys for future work.

AI-Augmented Business Process Management Systems (ABPMSs) are an emerging class of process-aware information systems, empowered by trustworthy AI technology. An ABPMS enhances the execution of business processes with the aim of making these processes more adaptable, proactive, explainable, and context-sensitive. This manifesto presents a vision for ABPMSs and discusses research challenges that need to be surmounted to realize this vision. To this end, we define the concept of ABPMS, we outline the lifecycle of processes within an ABPMS, we discuss core characteristics of an ABPMS, and we derive a set of challenges to realize systems with these characteristics., Comment: 19 pages, 1 figure

Numerous process discovery techniques exist for generating process models that describe recorded executions of business processes. The models are meant to generalize executions into human-understandable modeling patterns, notably parallelism, and enable rigorous analysis of process deviations. However, well-defined models with parallelism returned by existing techniques are often too complex or generalize the recorded behavior too strongly to be trusted in a practical business context. We bridge this gap by introducing the Probabilistic Inductive Miner (PIM) based on the Inductive Miner framework. PIM compares in each step the most probable operators and structures based on frequency information in the data, which results in block-structured models with significantly higher accuracy. All design choices in PIM are based on business context requirements obtained through a user study with industrial process mining experts. PIM is evaluated quantitatively and in an novel kind of empirical study comparing users' trust in discovered model structures. The evaluations show that PIM strikes a unique trade-off between model accuracy and model complexity, that is conclusively preferred by users over all state-of-the-art process discovery methods., accepted at IEEE International Conference on Process Mining (ICPM) 2021, submitted version

Business process management organizes work into several interrelated "units of work", fundamentally conceptualized as a task. The classical concept of a task as a single step executed by a single actor in a single case fails to capture more complex aspects of work that occur in real-life processes. For instance, actors working together or the processing of work in batches, where multiple actors and/or cases meet for a number of steps. Established process mining and modeling techniques lack concepts for dealing with these more complex manifestations of work. We leverage event graphs as a data structure to model behavior along the actor and the case perspective in an integrated model, revealing a variety of fundamentally different types of task executions. We contribute a novel taxonomy and interpretation of these task execution patterns as well as techniques for detecting these in event graphs, complementing recent research in identifying patterns of work and their changes in routine dynamics. Our evaluation on two real-life event logs shows that these non-classical task execution patterns not only exist, but make up for the larger share of events in a process and reveal changes in how actors do their work., Extended report underlying the paper "Classifying and Detecting Task Executions and Routines in Processes using Event Graphs" accepted at the Forum of the 19th International Conference on Business Process Management. 2021

Considerable amounts of data, including process events, are collected and stored by organisations nowadays. Discovering a process model from such event data and verification of the quality of discovered models are important steps in process mining. Many discovery techniques have been proposed, but none of them combines scalability with strong quality guarantees. We would like such techniques to handle billions of events or thousands of activities, to produce sound models (without deadlocks and other anomalies), and to guarantee that the underlying process can be rediscovered when sufficient information is available. In this paper, we introduce a framework for process discovery that ensures these properties while passing over the log only once and introduce three algorithms using the framework. To measure the quality of discovered models for such large logs, we introduce a model–model and model–log comparison framework that applies a divide-and-conquer strategy to measure recall, fitness, and precision. We experimentally show that these discovery and measuring techniques sacrifice little compared to other algorithms, while gaining the ability to cope with event logs of 100,000,000 traces and processes of 10,000 activities on a standard computer.

The detection of data breaches has become a major challenge for most organizations. The problem lies in the fact that organizations often lack proper mechanisms to control and monitor users' activities and their data usage. Although several auditing approaches have been proposed to assess the compliance of actual executed behavior, existing approaches focus on either checking data accesses against security policies (data perspective) or checking user activities against the activities needed to conduct business processes (process perspective). Analyzing user behavior from these perspectives independently may not be sufficient to expose security incidents. In particular, security incidents may remain undetected or diagnosed incorrectly. This paper proposes a novel auditing approach that reconciles the data and process perspectives, thus enabling the identification of a large range of deviations. In particular, we analyze and classify deviations with respect to the intended purpose of data and the context in which data are used, and provide a novel algorithm to identify non-conforming user behavior. The approach has been implemented in the open source framework ProM and was evaluated through both controlled experiments and a case study using real-life event data. The results show that the approach is able to accurately identify deviations in both data usage and control-flow, while providing the purpose and context of the identified deviations.

The Business Process Management field addresses design, improvement, management, support, and execution of business processes. In doing so, we argue that it focuses more on developing modeling notations and process design approaches than on the needs and preferences of the individual who is modeling (i.e., the user). New data-centric process modeling approaches are taken as a relevant and timely stream of process design approaches to test our argument. First, we provide a review of existing data-centric process approaches, culminating in a theoretical classification framework. Next, we empirically evaluate three specific approaches with regard to the claims they make. We had participants representative of actual users try out these approaches on realistic scenarios via a series of workshops. Participants assessed to what extent quality claims from the literature could be recognized within the workshop sessions. The results of this evaluation substantiate a number of claims behind the approaches, but also identify opportunities to further improve them. Most prominently, we found that the usability aspects of all considered approaches are a source of concern. This leads us to the insight that usability aspects of process design approaches are crucial and, in the perception of groups representative of actual users, leave much to be desired. In that sense, our research can be seen as a wake-up call for process modeling notation designers to consider the usability side—and as such, the interest of the human modeler—more than is currently the case.

Given a model of the expected behavior of a business process and given an event log recording its observed behavior, the problem of business process conformance checking is that of identifying and describing the differences between the process model and the event log. A desirable feature of a conformance checking technique is that it should identify a minimal yet complete set of differences. Existing conformance checking techniques that fulfill this property exhibit limited scalability when confronted to large and complex process models and event logs. One reason for this limitation is that existing techniques compare each execution trace in the log against the process model separately, without reusing computations made for one trace when processing subsequent traces. Yet, the execution traces of a business process typically share common fragments (e.g. prefixes and suffixes). A second reason is that these techniques do not integrate mechanisms to tackle the combinatorial state explosion inherent to process models with high levels of concurrency. This paper presents two techniques that address these sources of inefficiency. The first technique starts by transforming the process model and the event log into two automata. These automata are then compared based on a synchronized product, which is computed using an A* heuristic with an admissible heuristic function, thus guaranteeing that the resulting synchronized product captures all differences and is minimal in size. The synchronized product is then used to extract optimal (minimal-length) alignments between each trace of the log and the closest corresponding trace of the model. By representing the event log as a single automaton, this technique allows computations for shared prefixes and suffixes to be made only once. The second technique decomposes the process model into a set of automata, known as S-components, such that the product of these automata is equal to the automaton of the whole process model. A product automaton is computed for each S-component separately. The resulting product automata are then recomposed into a single product automaton capturing all the differences between the process model and the event log, but without minimality guarantees. An empirical evaluation using 40 real-life event logs shows that, used in tandem, the proposed techniques outperform state-of-the-art baselines in terms of execution times in a vast majority of cases, with improvements ranging from several-fold to one order of magnitude. Moreover, the decomposition-based technique leads to optimal trace alignments for the vast majority of datasets and close to optimal alignments for the remaining ones.

We propose and study dynamic versions of the classical flexibility constructs ‘skip’ and ‘block’ for workflows and motivate and define a formal semantics for them. We show that our semantics is a generalization of dead-path-elimination and solves the open problem to define dead-path-elimination for cyclic workflows. This in turn gives rise to a simple and fully local semantics for inclusive gateways. Finally, we show how our new constructs can be enacted on existing process engines by a workflow transformation that stores control-flow information in process variables.

Communication between organizations is formalized as process choreographies in daily business. While the correct ordering of exchanged messages can be modeled and enacted with current choreography techniques, no approach exists to describe and automate the exchange of data between processes in a choreography using messages. This paper describes an entirely model-driven approach for BPMN introducing a few concepts that suffice to model data retrieval, data transformation, message exchange, and correlation – four aspects of data exchange. For automation, this work utilizes a recent concept to enact data dependencies in internal processes. We present a modeling guideline to derive local process models from a given choreography; their operational semantics allows to correctly enact the entire choreography from the derived models only including the exchange of data. Targeting on successful interactions, we discuss means to ensure correct process choreography modeling. Finally, we implemented our approach by extending the camunda BPM platform with our approach and show its feasibility by realizing all service interaction patterns using only model-based concepts. Keywords: Process modeling; Data modeling; Process choreographies; Data exchange; BPMN; SQL

Workflow graphs represent the main control-flow constructs of industrial process modeling languages such as BPMN, EPC and UML Activity diagrams, whereas free-choice workflow nets is a well understood class of Petri nets that possesses many efficient analysis techniques. In this paper, we provide new results on the translation between workflow graphs and free-choice workflow nets. We distinguish workflow graphs with and without inclusive Or-logic. For workflow graphs without inclusive logic, we show that workflow graphs and free-choice workflow nets are essentially the same thing. More precisely, each workflow graph and each free-choice workflow net can be brought into an equivalent normal form such that the normal forms are, in some sense, isomorphic. This result gives rise to a translation from arbitrary free-choice workflow nets to workflow graphs. For workflow graphs with inclusive logic, we provide various techniques to replace inclusive Or-joins by subgraphs without inclusive logic, thus giving rise to translations from workflow graphs to free-choice nets. Additionally, we characterize the applicability of these replacements. Finally, we also display a simple workflow graph with an inclusive Or-join, which, in some sense, cannot be replaced. This shows a limitation of translating inclusive logic into free-choice nets and illustrates also a difficulty of translating inclusive logic into general Petri nets. Keywords: Workflow graphs; Petri nets; Free choice; Inclusive OR-join

In process mining, precision measures are used to quantify how much a process model overapproximates the behavior seen in an event log. Although several measures have been proposed throughout the years, no research has been done to validate whether these measures achieve the intended aim of quantifying over-approximation in a consistent way for all models and logs. This paper fills this gap by postulating a number of axioms for quantifying precision consistently for any log and any model. Further, we show through counter-examples that none of the existing measures consistently quantifies precision.

Process models discovered from a process log using process mining tend to be complex and have problems balancing between overfitting and underfitting. An overfitting model allows for too little behavior as it just permits the traces in the log and no other trace. An underfitting model allows for too much behavior as it permits traces that are significantly different from the behavior seen in the log. This paper presents a post-processing approach to simplify discovered process models while controlling the balance between overfitting and underfitting. The discovered process model, expressed in terms of a Petri net, is unfolded into a branching process using the event log. Subsequently, the resulting branching process is folded into a simpler process model capturing the desired behavior. Keywords: Process mining; Model simplification; Petri nets; Branching processes

Processes may require to execute the same activity in different stages of the process. A human modeler can express this by creating two different task nodes labeled with the same activity name (thus duplicating the task). However, as events in an event log often are labeled with the activity name, discovery algorithms that derive tasks based on labels only cannot discover models with duplicate labels rendering the results imprecise. For example, for a log where “payment” events occur at the beginning and the end of a process, a modeler would create two different “payment” tasks, whereas a discovery algorithm introduces a loop around a single “payment” task. In this paper, we present a general approach for refining labels of events based on their context in the event log as a preprocessing step. The refined log can be input for any discovery algorithm. The approach is implemented in ProM and was evaluated in a controlled setting. We were able to improve the quality of up to 42%of the models compared to using a log with imprecise labeling using default parameters and up to 87% using adaptive parameters. Moreover, using our refinement approach significantly increased the similarity of the discovered model to the original process with duplicate labels allowing for better rediscoverability. We also report on a case study conducted for a Dutch hospital.

We report on a case study on control-flow analysis of business process models. We checked 735 industrial business process models from financial services, telecommunications, and other domains. We investigated these models for soundness (absence of deadlock and lack of synchronization) using three different approaches: the business process verification tool Woflan, the Petri net model checker LoLA, and a recently developed technique based on SESE decomposition. We evaluate the various techniques used by these approaches in terms of their ability of accelerating the check. Our results show that industrial business process models can be checked in a few milliseconds, which enables tight integration of modeling with control-flow analysis. We also briefly compare the diagnostic information delivered by the different approaches and report some first insights from industrial applications. Keywords : Business Process Modeling; Quality assurance; Soundness; Verification; Control-flow analysis

Artifact-centric modeling is an approach for capturing business processes in terms of so-called business artifacts — key entities driving a company's operations and whose lifecycles and interactions define an overall business process. This approach has been shown to be especially suitable in the context of processes where one-to-many or many-to-many relations exist between the entities involved in the process. As a contribution towards building up a body of methods to support artifact-centric modeling, this article presents a method for automated discovery of artifact-centric process models starting from logs consisting of flat collections of event records. We decompose the problem in such a way that a wide range of existing (non-artifact-centric) automated process discovery methods can be reused in a flexible manner. The presented methods are implemented as a package for ProM, a generic open-source framework for process mining. The methods have been applied to reverse-engineer an artifact-centric process model starting from logs of a real-life business process.

One of the main challenges in process mining is to discover a process model describing observed behaviour in the best possible manner. Since event logs only contain example behaviour and one cannot assume to have seen all possible process executions, process discovery techniques need to be able to handle incompleteness. In this paper, we study the effects of such incomplete logs on process discovery. We analyse the impact of incompleteness of logs on behavioural relations, which are an abstraction often used by process discovery techniques. We introduce probabilistic behavioural relations that are less sensitive to incompleteness, and exploit these relations to provide a more robust process discovery algorithm. We prove this algorithm to be able to rediscover a model of the original system. Furthermore, we show in experiments that our approach even rediscovers models from incomplete event logs that are much smaller than required by other process discovery algorithms. Keywords: process discovery, block-structured process models, rediscoverability, process trees

Specification mining extracts candidate specification from existing systems, to be used for downstream tasks such as testing and verification. Specifically, we are interested in the extraction of behavior models from execution traces. In this paper we introduce mining of branching-time scenarios in the form of existential, conditional Live Sequence Charts, using a statistical data-mining algorithm. We show the power of branching scenarios to reveal alternative scenario-based behaviors, which could not be mined by previous approaches. The work contrasts and complements previous works on mining linear-time scenarios. An implementation and evaluation over execution trace sets recorded from several real-world applications shows the unique contribution of mining branching-time scenarios to the state-of-the-art in specification mining.

Process discovery is the problem of, given a log of observed behaviour, ¿nding a process model that ‘best’ describes this behaviour. A large variety of process discovery algorithms has been proposed. However, no existing algorithm guarantees to return a ¿tting model (i.e., able to reproduce all observed behaviour) that is sound (free of deadlocks and other anomalies) in ¿nite time. We present an extensible framework to discover from any given log a set of block-structured process models that are sound and ¿t the observed behaviour. In addition we characterise the minimal information required in the log to rediscover a particular process model. We then provide a polynomial-time algorithm for discovering a sound, ¿tting, block-structured model from any given log; we give suf¿cient conditions on the log for which our algorithm returns a model that is languageequivalent to the process model underlying the log, including unseen behaviour. The technique is implemented in a prototypical tool.

Scenario-based modeling is an approach for describing behaviors of a distributed system in terms of partial runs, called scenarios. Deriving an operational system from a set of scenarios is the main challenge that is typically addressed by either synthesizing system components or by providing operational semantics. Over the last years, several established scenario-based techniques have been adopted to Petri nets. Their adaptation allows for verifying scenario-based models and for synthesizing individual components from scenarios within one formal technique, by building on Petri net theory. However, current adaptations of scenarios face two limitations: a system modeler (1) cannot abstract from concrete behavior, and (2) cannot explicitly describe data in scenarios. This paper lifts these limitations for scenarios in the style of Live Sequence Charts (LSCs). We extend an existing model for scenarios, that features Petri net-based semantics, verification and synthesis techniques, and close the gap between LSCs and Petri nets further.

This report is the result of a Capita Selecta research project conducted at Eindhoven University of Technology. The objective of the project was to explore how batching behavior can be detected and measured in a quantitative manner from business process event logs.