van Bf Boudewijn Dongen

The prom framework: a new era in process mining tool support

Under the umbrella of buzzwords such as “Business Activity Monitoring” (BAM) and “Business Process Intelligence” (BPI) both academic (e.g., EMiT, Little Thumb, InWoLvE, Process Miner, and MinSoN) and commercial tools (e.g., ARIS PPM, HP BPI, and ILOG JViews) have been developed. The goal of these tools is to extract knowledge from event logs (e.g., transaction logs in an ERP system or audit trails in a WFM system), i.e., to do process mining. Unfortunately, tools use different formats for reading/storing log files and present their results in different ways. This makes it difficult to use different tools on the same data set and to compare the mining results. Furthermore, some of these tools implement concepts that can be very useful in the other tools but it is often difficult to combine tools. As a result, researchers working on new process mining techniques are forced to build a mining infrastructure from scratch or test their techniques in an isolated way, disconnected from any practical applications. To overcome these kind of problems, we have developed the ProM framework, i.e., an “pluggable” environment for process mining. The framework is flexible with respect to the input and output format, and is also open enough to allow for the easy reuse of code during the implementation of new process mining ideas. This paper introduces the ProM framework and gives an overview of the plug-ins that have been developed.

Process mining and verification of properties: an approach based on temporal logic

Information systems are facing conflicting requirements. On the one hand, systems need to be adaptive and self-managing to deal with rapidly changing circumstances. On the other hand, legislation such as the Sarbanes-Oxley Act, is putting increasing demands on monitoring activities and processes. As processes and systems become more flexible, both the need for, and the complexity of monitoring increases. Our earlier work on process mining has primarily focused on process discovery, i.e., automatically constructing models describing knowledge extracted from event logs. In this paper, we focus on a different problem complementing process discovery. Given an event log and some property, we want to verify whether the property holds. For this purpose we have developed a new language based on Linear Temporal Logic (LTL) and we combine this with a standard XML format to store event logs. Given an event log and an LTL property, our LTL Checker verifies whether the observed behavior matches the (un)expected/(un)desirable behavior.

Process mining : A two-step approach to balance between underfitting and overfitting

Process mining includes the automated discovery of processes from event logs. Based on observed events (e.g., activities being executed or messages being exchanged) a process model is constructed. One of the essential problems in process mining is that one cannot assume to have seen all possible behavior. At best, one has seen a representative subset. Therefore, classical synthesis techniques are not suitable as they aim at finding a model that is able to exactly reproduce the log. Existing process mining techniques try to avoid such “overfitting” by generalizing the model to allow for more behavior. This generalization is often driven by the representation language and very crude assumptions about completeness. As a result, parts of the model are “overfitting” (allow only for what has actually been observed) while other parts may be “underfitting” (allow for much more behavior without strong support for it). None of the existing techniques enables the user to control the balance between “overfitting” and “underfitting”. To address this, we propose a two-step approach. First, using a configurable approach, a transition system is constructed. Then, using the “theory of regions”, the model is synthesized. The approach has been implemented in the context of ProM and overcomes many of the limitations of traditional approaches.

Detection and prediction of errors in EPCs of the SAP reference model

Up to now there is neither data available on how many errors can be expected in process model collections, nor is it understood why errors are introduced. In this article, we provide empirical evidence for these questions based on the SAP reference model. This model collection contains about 600 process models expressed as Event-driven Process Chains (EPCs). We translated these EPCs into YAWL models, and analyzed them using the verification tool WofYAWL. We discovered that at least 34 of these EPCs contain errors. Moreover, we used logistic regression to show that complexity of EPCs has a significant impact on error probability.

Conformance Checking Using Cost-Based Fitness Analysis

The growing complexity of processes in many organizations stimulates the adoption of business process analysis techniques. Typically, such techniques are based on process models and assume that the operational processes in reality conform to these models. However, experience shows that reality often deviates from hand-made models. Therefore, the problem of checking to what extent the operational process conforms to the process model is important for process management, process improvement, and compliance. In this paper, we present a robust replay analysis technique that is able to measure the conformance of an event log for a given process model. The approach quantifies conformance and provides intuitive diagnostics (skipped and inserted activities). Our technique has been implemented in the ProM 6framework. Comparative evaluations show that the approach overcomes many of the limitations of existing conformance checking techniques.

Multi-phase Process Mining: Building Instance Graphs

Deploying process-driven information systems is a time-con-suming and error-prone task. Process mining attempts to improve this by automatically generating a process model from event-based data. Existing techniques try to generate a complete process model from the data acquired. However, unless this model is the ultimate goal of mining, such a model is not always required. Instead, a good visualization of each individual process instance can be enough. From these individual instances, an overall model can then be generated if required. In this paper, we present an approach which constructs an instance graph for each individual process instance, based on information in the entire data set. The results are represented in terms of Event-driven Process Chains (EPCs). This representation is used to connect our process mining to a widely used commercial tool for the visualization and analysis of instance EPCs.

Towards Robust Conformance Checking

The growing complexity of processes in many organizations stimulates the adoption of business process management (BPM) techniques. Process models typically lie at the basis of these techniques and generally, the assumption is made that the operational business processes as they are taking place in practice conform to these models. However, recent experience has shown that this often isn’t the case. Therefore, the problem of checking to what extent the operational process conforms to the process model is increasingly important.

Verification of the SAP reference models using EPC reduction, state-space analysis, and invariants

A reference model is a generic conceptual model that formalizes recommended practices for a certain domain. Today, the SAP reference models are among the most comprehensive reference models, including over 4000 entity types and covering over 1000 business processes and inter-organizational scenarios. The SAP reference models use Event-driven Process Chains (EPCs) to model these processes and scenarios. Like other informal languages, EPCs are intended to support the transition from a business model to an executable model. For this reason, researchers have tried to formalize the semantics of EPCs. However, in their approaches, they fail to acknowledge the fact that in EPCs constructs exist that require human judgment to assess correctness. This paper aims to acknowledge this fact by introducing a two-step approach. First, the EPC is reduced using universally accepted reduction rules. Second, the reduced EPC is analyzed using a mixture of state-space analysis, invariants, and human judgment. This approach has been implemented in a tool, and applying this tool to the SAP reference models showed that these contain errors, which clearly shows the added value of this verification approach.

A genetic algorithm for discovering process trees

Existing process discovery approaches have problems dealing with competing quality dimensions (fitness, simplicity, generalization, and precision) and may produce anomalous process models (e.g., deadlocking models). In this paper we propose a new genetic process mining algorithm that discovers process models from event logs. The tree representation ensures the soundness of the model. Moreover, as experiments show, it is possible to balance the different quality dimensions. Our genetic process mining algorithm is the first algorithm where the search process can be guided by preferences of the user while ensuring correctness.

Faulty EPCs in the SAP reference model

Little is known about error probability in enterprise models as they are usually kept private. The SAP reference model is a publically available model that contains more than 600 non-trivial process models expressed in terms of Event-driven Process Chains (EPCs). We have automatically translated these EPCs into YAWL models and analyzed these models using WofYAWL, a verification tool based on Petri nets, in order to acquire knowledge about errors in large enterprise models. We discovered that at least 34 of these EPCs contain errors (i.e., at least 5.6% is flawed) and analyzed which parts of the SAP reference model contain most errors. This systematic analysis of the SAP reference model illustrates the need for verification tools such as WofYAWL.