Workflow mining with InWoLvE

doi:10.1016/j.compind.2003.10.002

Computers in Industry

Volume 53, Issue 3, April 2004, Pages 245-264

https://doi.org/10.1016/j.compind.2003.10.002 Get rights and content

Abstract

State of the art information systems are based on explicit process models called workflow models. Experience from industrial practice shows that the definition of workflow models is a very time consuming and error prone task. Recently, there has been an increasing interest in applying techniques from data mining and machine learning to support this task. This approach has also been termed as process or workflow mining. In this paper, we give an overview of the algorithms that were implemented within the InWoLvE workflow mining system, we summarize the most important results of their experimental evaluation and we present the experiences that were made in the first industrial application of InWoLvE.

Introduction

State of the art information systems are based on explicit process models called workflow models. These models are interpreted by one or more workflow engines to drive the execution of business processes within or across several enterprises. Experience from industrial practice shows that the definition of workflow models is a very time consuming and error prone task. In depth knowledge of the business process and the ability to represent this knowledge using a formal workflow modelling language are needed for this task. Recently, there has been an increasing interest in applying techniques from data mining and machine learning to support this task [1], [2], [3], [4], [5], [6], [7], [8], [9]. This approach has also been termed as process or workflow mining. The basic idea of the workflow mining approach is to collect traces of workflow executions and to derive a workflow model from these observations. This is useful for example if some information system supporting the process, that logs all relevant events, is already in place before the workflow model is defined. Furthermore workflow mining techniques and advanced workflow technology, which is moving towards more operational flexibility [10], [11], [12], [13], enable an evolutionary approach to the development of workflow applications, where an initially roughly defined and informal or semi-formal workflow model is iteratively refined and formalized.

In this paper, we give an overview of the algorithms that were implemented within the InWoLvE workflow mining system, we summarize the most important results of their experimental evaluation and we present the experiences that were made in the first industrial application of InWoLvE. The remainder of this paper is organized as follows. Section 2 defines the most important terms used throughout this paper. Section 3 formalizes the workflow mining problem, it defines problem classes and it gives an overview of the induction and transformation algorithms used within the workflow mining system InWoLvE. In Section 4, we describe the InWoLvE prototype and we summarize the most important results of its experimental evaluation. The experiences we have made in the first industrial application of InWoLvE are covered by Section 5. In Section 6, we discuss related work and finally in Section 7 we summarize the main conclusions and give an outlook on our future work.

Section snippets

Activities

The basic element of workflow instances, workflow models and stochastic activity graphs are activities. In the following, we use A={a_△,a₀,a₁,…,a_n} for the set of all activities. The activities a_△ and a₀ denote special invisible activities that mark the beginning and the end of a model, an instance or a stochastic activity graph (SAG).

Workflow instance

Definition 2.1 (Workflow instance)

A workflow instance is a tuple e=(K_e, <_e, f_e, k_△, k₀), where

•
K_e={k_△,k₀,k₁,…k_{n_e}} is a set of nodes (activity instances),
•
<_e⊆(K_e×K_e) is a partial order on K_e,
•
f_e:K_e→A

The workflow mining algorithm

The workflow mining task we try to solve can be described as follows: given a set of workflow instances E, find a good approximation W of the workflow model W₀, that generated E. Of course W₀ need not exist. It is simply a modeling hypothesis.

In this contribution, we focus on the induction of the workflow structure. Methods for mining transition conditions are discussed in [7], [15]. Our workflow mining algorithm solves the structure mining task in two steps:

•
In the first step of our workflow

The InWoLvE prototype

The workflow mining tool InWoLvE (Inductive Workflow Learning via Examples)⁵ implements the workflow mining algorithm splitPar from Section 3.2 and the transformation algorithm SAGtoADL from Section 3.3. Furthermore two induction algorithms restricted to sequential workflow models have been implemented. The first one

An application to product development processes

In the following two subsections we describe the application of InWoLvE to two real-world processes. Both processes are subprocesses of the product development process for Mercedes Benz passenger cars. The first process is a slightly simplified version of the part release process and the second process is a software change management process for electronic control units (ECUs). In the first case study we applied InWoLvE to an example set generated by simulation as in the experiments described

Related work

In [2] an approach called process mining, based on the induction of directed graphs, is presented. The main deficits compared to our approach are that the algorithm by Agrawal et. al. is restricted to workflow models of problem class 3 and that the nature of splits and joins is not analyzed and explicitly described by the generated workflow model. This makes it much more difficult for process experts to interpret the result. Furthermore, it is not guaranteed that the generated model is

Conclusions and future work

We have described the most important algorithms that were implemented in the workflow mining system InWoLvE. InWoLvE solves the workflow mining task in two steps. In the first step it creates a stochastic activity graph from the example set and in the second step it transforms this stochastic activity graph into a well-defined workflow model. The experiments we have described show that InWoLvE is applicable for a wide range of workflow models. We have also presented the experiences made during

Joachim Herbst studied computer science at the University of Ulm, where he also did his PhD in the area of workflow mining. Since 1995, he has been working for DaimlerChrysler Research and Technology. His research interests include machine learning, workflow management, enterprise application integration and concurrent engineering.

References (24)

W.V. Aalst, A. Weijters, L. Maruster, Workflow mining: which processes can be rediscovered? Beta working paper series,...
R. Agrawal, D. Gunopulos, F. Leymann, Mining process models from workflow logs, in: Proceedings of the Sixth...
J. Cook et al.
Discovering models of software processes from event-based data
ACM Transactions on Software Engineering and Methodology
(1998)
J. Cook, A. Wolf, Event-Based Detection of Concurrency, Technical Report, CU-CS-860-98, Department of Computer Science,...
J. Herbst, D. Karagiannis, Integrating machine learning and workflow management to support acquisition and adaptation...
J. Herbst, A machine learning approach to workflow management, in: Proceedings of the 11th European Conference on...
J. Herbst, Ein induktiver Ansatz zur Akquisition und Adaption von Workflow-Modellen, Ph.D. Thesis, Universität Ulm,...
G. Schimm, Process mining elektronischer Geschäftsprozesse, in: Proceedings Elektronische Geschäftsprozesse,...
A. Weijters, W.V. Aalst, Rediscovering workflow models from event-based data, in: Proceedings of the 11th Dutch-Belgian...
W.V. Aalst, J. Desel, A. Oberweis, Business Process Management: Models, Techniques, and Empirical Studies, Springer,...

M. Klein, C. Dellarocas, A. Bernstein, Proceedings of the CSCW-98 Workshop Towards Adaptive Workflow Systems, Seattle,...

M. Klein et al.

Introduction to the special issue on adaptive workflow systems

Computer Supported Cooperative Work

(2000)

Cited by (123)

Formal approach for discovering work transference networks from workflow logs
2020, Information Sciences
Citation Excerpt :
This makes the workflow mining and discovery functionality, to reflect these paradigmatic properties, essential features of workflow management systems. In terms of collecting, filtering, and discovering activities, using workflow logs and warehouses, the workflow literature to date has been focused mainly on two types of workflow discovery activities: workflow process discovery [10,17–20] and workflow knowledge discovery [2,3,8,21–26]. Workflow process discovery is directly concerned with redesigning and reengineering the process aspect (control flow) of workflow models and applications by rediscovering temporal workcases from the workflow logs and warehouses, whereas workflow knowledge discovery is closely connected with replanning and realigning the organizational resource aspect (such as data flow, performer, role, and program) of workflow models and applications by rediscovering enacting and binding histories (e.g., temporal worktransferences) from the workflow logs and warehouses.
This paper proposes a formal principle for discovering work transference networks of workflow-supported organizational employees from workflow enactment histories contained in event logs. This originates from the strong belief that those work transference networks, hidden in workflow enactment activities and histories, can both connote the degree of work sharing and work relevancy between workflow performers and denote their degree of work intensity. In this paper, we devise a series of formal definitions and algorithms for discovering a work transference network from a workflow procedure, and from its enactment histories in event logs. The final goal of the paper is to theoretically build the principle of fidelity into workflow human resource planning and its performance, via a novel concept of work transference networks that can be discovered from a workflow model and, moreover, rediscovered from its enactment history. In deploying the proposed principle, we base the formal representation on information control net theory, which can graphically and mathematically represent workflow procedures. We apply directed graph theory to formally and graphically define the work transference network model proposed in this paper. For the sake of verifying and validating the proposed concepts and algorithms, we implement a work transference network rediscovery system, and apply it to a workflow enactment event log dataset, in an experimental study. Finally, we describe the implications of discovering and rediscovering work transference networks, as a human resource management and evaluation technique, in workflow-supported enterprises and organizations.
Process mining techniques and applications – A systematic mapping study
2019, Expert Systems with Applications
Process mining is a growing and promising study area focused on understanding processes and to help capture the more significant findings during real execution rather than, those methods that, only observed idealized process model. The objective of this article is to map the active research topics of process mining and their main publishers by country, periodicals, and conferences. We also extract the reported application studies and classify these by exploration domains or industry segments that are taking advantage of this technique. The applied research method was systematic mapping, which began with 3713 articles. After applying the exclusion criteria, 1278 articles were selected for review. In this article, an overview regarding process mining is presented, the main research topics are identified, followed by identification of the most applied process mining algorithms, and finally application domains among different business segments are reported on. It is possible to observe that the most active research topics are associated with the process discovery algorithms, followed by conformance checking, and architecture and tools improvements. In application domains, the segments with major case studies are healthcare followed by information and communication technology, manufacturing, education, finance, and logistics.
Enhancing discovered processes with duplicate tasks
2016, Information Sciences
Citation Excerpt :
Therefore, these algorithms, when dealing with logs with duplicate tasks, usually give as a result models with overly connected nodes or needless loops, decreasing the precision and simplicity of the model. On the other hand, there are techniques that do not make such a restrictive assumption [3,6–9,13–15,19]. Typically, all these techniques identify the potential duplicate activities in a pre-mining step, or during the mining process.
Including duplicate tasks in the mining process is a challenge that hinders the process discovery, as it is also necessary to find out which events of the log belong to which transitions. To face this problem, we propose SLAD (Splitting Labels After Discovery), an algorithm that uses the local information of the log to enhance an already mined model, by performing a local search over the tasks that have more probability to be duplicated in the log. This proposal has been validated with 54 different mined models from three process discovery algorithms, improving the final solution in 45 of the cases. Furthermore, SLAD has been tested in a real scenario.
Method of Webpage Entity Extraction Based on Mixed Attribute Measurement and DOM Tree
2022, ACM International Conference Proceeding Series
A novel method for discovering process based on the network analysis approach in the context of social commerce systems
2021, Journal of Theoretical and Applied Electronic Commerce Research
Practical Aspects for Exploration and Analysis of Manual Interventions in Process Plants
2021, IEEE International Conference on Industrial Informatics (INDIN)

View all citing articles on Scopus

Dimitris Karagiannis studied Computer Science at the Technical University of Berlin and graduated several visit stays in the USA and Japan. From 1987 until 1992, he was business unit manager for Business Information Sytems at the Research Institute for Applied knowledge Management (FAW) in Ulm. In 1993, he founded the Department of Knowledge Engineering at the Insitute for Computer Science and Business Informatics at the University of Vienna, focusing on knowledge management, business intelligence and meta-modelling. Prof. Karagiannis has published lot of scientifical research papers in the field of databases, expert systems, business process management, workflow-systems and knowledge management. He is the author of two books concerned with Knowledge Databases and Knowledge Management and is engaged in national and EU-funded research projects. The Business Process Management Approach he established, which is concerned with the thematic of Knowledge- and Business Process Management, has been succesfully implemented in several service companies. He founded the european software- and consulting company BOC ITC Ltd. (http://www.boc-eu.com), which realised the development and implementation of the business process management toolkit ADONIS^®.

View full text