Business Process Management

Analysing performance of business processes is an important vehicle to improve their operation. Specifically, an accurate assessment of sojourn times and remaining times enables bottleneck analysis and resource planning. Recently, methods to create respective performance models from event logs have been proposed. These works are severely limited, though: They either consider control-flow and performance information separately, or rely on an ad-hoc selection of temporal relations between events. In this paper, we introduce the Temporal Network Representation (TNR) of a log, based on Allen’s interval algebra, as a complete temporal representation of a log, which enables simultaneous discovery of control-flow and performance information. We demonstrate the usefulness of the TNR for detecting (unrecorded) delays and for probabilistic mining of variants when modelling the performance of a process. In order to compare different models from the performance perspective, we develop a framework for measuring performance fitness. Under this framework, we provide guarantees that TNR-based process discovery dominates existing techniques in measuring performance characteristics of a process. To illustrate the practical value of the TNR, we evaluate the approach against three real-life datasets. Our experiments show that the TNR yields an improvement in performance fitness over state-of-the-art algorithms.

Synthesis aims at producing a process model from specified sample executions. A user can specify a set of executions of a system in a specification language that is much simpler than a process modeling language. The intended process model is then constructed automatically.

Synthesis algorithms have been extensively explored for cases where the specification language is a reachability graph or a sequential language. Concerning synthesis from partial languages, however, there is a significant gap between theory and practical application. In the literature, we find two different synthesis methods for partial languages, but both have poor runtime even in reasonably sized practical examples. In this paper, we introduce a new and more efficient synthesis algorithm for partial languages based on Hasse diagrams.

Privacy Enhancing Technologies (PETs) play an important role in preventing privacy leakage of data along information flows. Although business process modelling is well-suited for expressing stakeholder collaboration and process support by technical solutions, little is done to visualise and analyse privacy leakages in the processes. We propose PE-BPMN – privacy-enhanced extensions to the BPMN language for capturing data leakages. We demonstrate its feasibility in the mobile app scenario where private data leakages are determined. Our approach helps system builders make decisions on the privacy solutions at the early stages of development and lets auditors analyse existing systems.

Process discovery techniques return process models that are either formal (precisely describing the possible behaviors) or informal (merely a “picture” not allowing for any form of formal reasoning). Formal models are able to classify traces (i.e., sequences of events) as fitting or non-fitting. Most process mining approaches described in the literature produce such models. This is in stark contrast with the over 25 available commercial process mining tools that only discover informal process models that remain deliberately vague on the precise set of possible traces. There are two main reasons why vendors resort to such models: scalability and simplicity. In this paper, we propose to combine the best of both worlds: discovering hybrid process models that have formal and informal elements. As a proof of concept we present a discovery technique based on hybrid Petri nets. These models allow for formal reasoning, but also reveal information that cannot be captured in mainstream formal models. A novel discovery algorithm returning hybrid Petri nets has been implemented in ProM and has been applied to several real-life event logs. The results clearly demonstrate the advantages of remaining “vague” when there is not enough “evidence” in the data or standard modeling constructs do not “fit”. Moreover, the approach is scalable enough to be incorporated in industrial-strength process mining tools.

Processes control critical IT systems and business cases in dynamic environments. Hence, ensuring secure model executions is crucial to prevent misuse and attacks. In general, anomaly detection approaches can be employed to tackle this challenge. Existing ones analyze each process instance individually. Doing so does not consider attacks that combine multiple instances, e.g., by splitting fraudulent fund transactions into multiple instances with smaller “unsuspicious” amounts. The proposed approach aims at detecting such attacks. For this, anomalies between the temporal behavior of a set of historic instances (ex post) and the temporal behavior of running instances are identified. Here, temporal behavior refers to the temporal order between the instances and their events. The proposed approach is implemented and evaluated based on real life process logs from different domains and artificial anomalies.

Event logs of a self-care troubleshooting portal indicate that most customers do not follow the directions up to a conclusive end. Consequently, customers risk losing confidence in the support channel, which undermines the competitive strength of the business. We present a method for visual analysis of the event logs that employs graph reduction, and the use of path classification to create a novel type of flow diagram. These diagrams help to discover and communicate new insights, such as important trends about the way the customer traverses through the underlying business process.

A fundamental assumption of Business Process Management (BPM) is that redesign delivers new and improved versions of business processes. This assumption, however, does not necessarily hold, and required compensatory action may be delayed until a new round in the BPM life-cycle completes. Current approaches to process redesign face this problem in one way or another, which makes rapid process improvement a central research problem of BPM today. In this paper, we address this problem by integrating concepts from process execution with ideas from DevOps. More specifically, we develop a technique called AB-BPM that offers AB testing for process versions with immediate feedback at runtime. We implemented this technique in such a way that two versions (A and B) are operational in parallel and any new process instance is routed to one of them. The routing decision is made at runtime on the basis of the achieved results for the registered performance metrics of each version. AB-BPM provides for ultimate convergence towards the best performing version, no matter if it is the old or the new version. We demonstrate the efficacy of our technique by conducting an extensive evaluation based on both synthetic and real-life data.

Blockchain technology enables the execution of collaborative business processes involving untrusted parties without requiring a central authority. Specifically, a process model comprising tasks performed by multiple parties can be coordinated via smart contracts operating on the blockchain. The consensus mechanism governing the blockchain thereby guarantees that the process model is followed by each party. However, the cost required for blockchain use is highly dependent on the volume of data recorded and the frequency of data updates by smart contracts. This paper proposes an optimized method for executing business processes on top of commodity blockchain technology. Our optimization targets three areas specifically: initialization cost for process instances, task execution cost by means of a space-optimized data structure, and improved runtime components for maximized throughput. The method is empirically compared to a previously proposed baseline by replaying execution logs and measuring resource consumption and throughput.

As for all kind of software, customers expect to find business process execution provided as a service (BPMaaS). They expect it to be provided at the best cost with guaranteed SLA. From the BPMaaS provider point of view it can be done thanks to the provision of an elastic cloud infrastructure. Providers still have to provide the service at the lowest possible cost while meeting customers expectation. We propose a customer-centric service model that link the BP execution requirement to cloud resources, and that optimize the deployment of customer’s (or tenants) processes in the cloud to adjust constantly the provision to the needs. However, migrations between cloud configurations can be costly in terms of quality of service and a provider should reduce the number of migrations. We propose a model for BPMaaS cost optimization that take into account a maximum number of migrations for each tenants. We designed a heuristic algorithm and experimented using various customer load configurations based on customer data, and on an actual estimation of the capacity of cloud resources.

The monitoring of project-oriented business processes is difficult because their state is fragmented and represented by the progress of different documents and artifacts being worked on. This observation holds in particular for software development projects in which various developers work on different parts of the software concurrently. Prior contributions in this area have proposed a plethora of techniques to analyze and visualize the current state of the software artifact as a product. It is surprising that these techniques are missing to provide insights into what types of work are conducted at different stages of the project and how they are dependent upon another. In this paper, we address this research gap and present a technique for mining the software process including dependencies between artifacts. Our evaluation of various open-source projects demonstrates the applicability of our technique.

The interest of integrating decision analysis approaches with the automated discovery of processes from data has seen a vast surge over the past few years. Most notably the introduction of the Decision Model and Notation (DMN) standard by the Object Management Group has provided a suitable solution for filling the void of decision representation in business process modeling languages. Process discovery has already embraced DMN for so-called decision mining, however, the efforts are still limited to a control flow point of view, i.e., explaining routing (constructs) or decision points. This work, however, introduces an integrated way of capturing the decisions that are embedded in the process, which is not limited to local characteristics, but provides a decision model in the form of a decision diagram which encompasses the full process execution span. Therefore, a typology is proposed for classifying different activities that contribute to the decision dimension of the process. This enables the possibility for an in-depth analysis of every activity, deciding whether it entails a decision, and what its relation is to other activities. The findings are implemented and illustrated on the 2013 BPI Challenge log, an exemplary dataset originating from a decision-driven process.

Business process models are widely used in organizations by information systems analysts to represent complex business requirements and by business users to understand business operations and constraints. This understanding is extracted from graphical process models as well as business rules. Prior research advocated integrating business rules into business process models to improve the effectiveness of important organizational activities, such as developing shared understanding, effective communication, and process improvement. However, whether such integrated modeling can improve the understanding of business processes has not been empirically evaluated. In this paper, we report on an experiment that investigates the effect of linked rules, a specific rule integration approach, on business process model understanding. Our results indicate that linked rules are associated with better time efficiency in interpreting business operations, less mental effort, and partially associated with improved accuracy of understanding.

Business process models can serve different purposes, from discussion and analysis among stakeholders, to simulation and execution. While work has been done on deriving modeling guidelines to improve understandability, it remains to be determined how different modeling practices impact the execution of the models. In this paper we observe how semantically equivalent, but syntactically different, models behave in order to assess the performance impact of different modeling practices. To do so, we propose a methodology for systematically deriving semantically equivalent models by applying a set of model transformation rules and for precisely measuring their execution performance. We apply the methodology on three scenarios to systematically explore the performance variability of 16 different versions of parallel, exclusive, and inclusive control flows. Our experiments with two open-source business process management systems measure the execution duration of each model’s instances. The results reveal statistically different execution performance when applying different modeling practices without total ordering of performance ranks.

A workflow (WF) is a formal description of a business process in which single atomic work units (tasks), organized in a partial order, are assigned to processing entities (agents) in order to achieve some business goal(s). A workflow management system must coordinate the execution of tasks and WF instances. Usually, the assignment of tasks to agents is accomplished by external constraints not represented in a WF. An access-controlled workflow (ACWF) extends a classical WF by explicitly representing agent availability for each task and authorization constraint. Authorization constraints model which users are authorized for which tasks depending on “who did what”. Recent research has addressed temporal controllability of WFs under conditional and temporal uncertainty. However, controllability analysis for ACWFs under conditional uncertainty has never been addressed before. In this paper, we define weak, strong and dynamic controllability of ACWFs under conditional uncertainty, we present algorithmic approaches to address each of these types of controllability, and we synthesize execution strategies that specify which user has been (or will be) assigned to which task.

Predictive business process monitoring aims at leveraging past process execution data to predict how ongoing (uncompleted) process executions will unfold up to their completion. Nevertheless, cases exist in which, together with past execution data, some additional knowledge (a-priori knowledge) about how a process execution will develop in the future is available. This knowledge about the future can be leveraged for improving the quality of the predictions of events that are currently unknown. In this paper, we present two techniques - based on Recurrent Neural Networks with Long Short-Term Memory (LSTM) cells - able to leverage knowledge about the structure of the process execution traces as well as a-priori knowledge about how they will unfold in the future for predicting the sequence of future activities of ongoing process executions. The results obtained by applying these techniques on six real-life logs show an improvement in terms of accuracy over a plain LSTM-based baseline.

Knowledge-intensive Processes (KiPs) are unstructured processes that demand an understanding beyond control flow and data. Being knowledge-centric and varying at each instance, KiPs demand new perspectives for proper process analysis. Most KiPs have strong collaboration characteristics, where interactions among participants are crucial to achieve process goals. Process participants perform activities and collaborate with each other, driven by their Beliefs, Desires and Intentions; therefore, the analysis of these elements is vital to the correct understanding, modeling and execution of a KiP. This research proposes a method based on Speech Act Theory and Process Mining to discover the flow of speech acts related to Beliefs, Desires and Intentions from event logs, and shows how this relation fosters process performance analysis. The approach was evaluated through a case study in a real life scenario, and results showed that relevant insights in forms of speech acts flow patterns were discovered and related to performance issues of the KiP.

Business processes are dynamic and change due to diverse factors. While existing approaches aim to detect drifts in the process structure, TESSERACT looks for temporal drifts in activity interim times. This orthogonal view on the process extends the traditional data cube of events - case id, activities and timestamps - by a fourth dimension and improves the operational support by a visualization of temporal drifts in real-time.

Insights about temporal deviations lead to an augmented awareness of imminent failures or improved service times. The detection of related structural concept drifts can be improved by early warning, as operation times of critical parts often increase before they catastrophically fail.

Process mining has focused, among others, on the discovery of frequent behavior with the aim to understand what is mainly happening in a process. Little work has been done involving uncommon behavior, and mostly centered on the detection of anomalies or deviations. But infrequent behavior can be also important for the management of a process, as it can reveal, for instance, an uncommon wrong realization of a part of the process. In this paper, we present WoMine-i, a novel algorithm to retrieve infrequent behavioral patterns from a process model. Our approach searches in a process model extracting structures with sequences, selections, parallels and loops, which are infrequently executed in the logs. This proposal has been validated with a set of synthetic and real process models, and compared with state of the art techniques. Experiments show that WoMine-i can find all types of patterns, extracting information that cannot be mined with the state of the art techniques.