Simplifying process model abstraction: Techniques for generating model names
Introduction
Recent years have seen a substantial increase in business process modeling initiatives. While process modeling was utilized in the 1990s mainly as a technique for facilitating single process re-engineering efforts [1], [2], [3], [4], [5], many companies have turned to a more encompassing and evolutionary approach to Business Process Management (BPM). This development has led to the establishment of BPM expert teams, competence centers, and consulting departments within organizations. These units typically manage a central repository of business process models capturing various aspects of an organization's business operations. Process modeling is an ongoing activity in a setting that is interwoven with strategic management, quality assurance, controlling, and the specific functional areas of a company. As a result, companies that model business processes often maintain repositories containing hundreds if not thousands of models at a remarkable level of detail [6].
The mass of documentation stored in a process model repository poses considerable challenges for efficient and effective use of these models. At various stages, different stakeholders require individual views on these models for getting an overview, receiving support during their modeling efforts, and for quality assurance purposes. For example, abstract views are a good vehicle for achieving a cognitive fit between the representation of the process and the task at hand, cf. [7], [8]. Moreover, they offer a suitable aggregation of process information, which has been found to be crucial for process improvement initiatives [9].
A survey on how process models are used in practice reveals no less than fifteen different use cases for abstracting and compressing a detailed, fine-granular model toward a smaller model capturing the essential information a stakeholder is after [10]. These use cases clearly motivate the development of techniques that facilitate such transformations, e.g., [11], [12], [13]. These techniques, however, do not address the problem of how to automatically generate a proper name for an abstract model. This puts the burden on the stakeholder to come up with a meaningful reference for each new abstract model that is derived. This is a task that cannot be ignored, since much of the meaning of a process model can only be derived from what its elements stand for. It is also potentially a highly repetitive task: abstractions for a single process model may be dynamically generated requiring a name proposal each time they are viewed by a stakeholder. The same problem occurs when searching for similar or identical model fragments in large process model repositories. Such “(approximate) clones” can be automatically found [14], [15], [16], [17], but then need to be stored with a meaningful name. An automatic technique for determining the name of a process model clone is not yet available, thus hindering the wider adoption of clone detection in practice. In this paper, we address the problem of defining names for process models and process model fragments that, for instance, result from applying abstraction or clone identification. Our contribution is an approach that will ease the naming task for users who are interested in creating more abstract views on process models than are readily available to them. Specifically, our approach builds on techniques that analyze the activity labels of a process model, and returns a ranked list of names from which the top ranked name is provided. To this end, we build on insights from theories of meaning and exploratory research to devise naming strategies for process models. As we will argue, the quality of the generated naming suggestions is comparable to those that would be generated by humans; however, in the proposed approach these suggestions are generated in only a fraction of the time a human modeler would need to inspect the underlying (fine-grained) model elements and their interrelations.
We implemented the naming approach in a prototype tool based on natural language processing techniques, and, in order to demonstrate the flexibility of our approach, we configured the tool to detect names for process models specified in two languages: English and German. Next, we validated the approach in various directions. First, we measured the performance and accuracy of our approach using three large datasets from practice. Second, we conducted a case study involving process modelers with varying expertise to evaluate the usefulness and appropriateness of the approach in practice.
The rest of the paper is structured as follows. Section 2 reviews the essential concepts of business process modeling, presents a taxonomy of naming strategies grounded in theories of meaning, and highlights an exploratory study on process model names. Section 3 introduces the approach for automatically deriving process model names. Section 4 discusses the results of the evaluation while Section 5 summarizes achievements, and suggests implications for research and practice. Section 6 concludes the paper with an outlook on future research.
Section snippets
Background
In this section, we discuss background work required for our research. First, we present the essential aspects of business process modeling. Next, we identify different techniques for finding a process model name. These techniques form the basis for the approach that is proposed in this paper.
Derivation of name proposal
The different theories of meaning exhibit relative strengths and weaknesses. Semantic theories are typically subject to a context of utterance reflecting the circumstances of evaluation, and to modes of presentation. Foundational theories emphasize the importance of: the speakers’ meaning and intentions, beliefs, and social norms. Thus, names of process models will have different meanings in different settings, just as the process of registering a new-born child will hardly be exactly the same
Evaluation
To demonstrate the capability of our approach to find appropriate process model names, we conducted a two-part evaluation. First, we conducted an experiment using our prototype implementation and different process model collections from practice in order to demonstrate the efficiency of the implementation and the semantic closeness of the proposed model names to the original names. Second, we ran a case study with a large insurance company to reflect on the usefulness and appropriateness of the
Implications
In this section we discuss the implications of our research for research and practice.
Conclusion
In this paper, we addressed the problem of automatically finding suitable process model names. Based on insights from theories of meaning, we defined a novel approach for proposing a name for a business process model. This approach takes the labels of activities and events of the model into account, and does not depend upon external knowledge provided by the user. We implemented the approach in a prototype tool and evaluated it with different sets of process models from practice. The results
References (89)
- et al.
Fast detection of exact clones in business process model repositories
Information Systems
(2013) Family resemblancesstudies in the internal structure of categories
Cognitive Psychology
(1975)Deriving verbs in English
Language Sciences
(2008)- et al.
Jan mendling: on the refactoring of activity labels in business process models
Information Systems (IS)
(2012) - et al.
Activity labeling in process modelingempirical insights and recommendations
Information Systems
(2010) - et al.
Process instantiation
Data & Knowledge Engineering
(2009) - et al.
Workflow modeling for virtual processes: an order-preserving process-view approach
Information Systems
(2003) - et al.
Syntax highlighting in business process models
Decision Support Systems
(2011) - et al.
View integration of behavior in object-oriented databases
Data & Knowledge Engineering
(2001) - et al.
Similarity of business process modelsmetrics and evaluation
Information Systems
(2011)
Communication flow orientation in business process modeling and its effect on redesign successresults from a field study
Decision Support Systems
Seven process modeling guidelines (7PMG)
Information and Software Technology
Managing large collections of business process models – current techniques and challenges
Computers in Industry
The new industrial engineering: information technology and business process redesign
Sloan Management Review
Reengineering workdon't automate, obliterate
Harvard Business Review
Strategies for business process reengineeringevidence from field studies
Journal of Management Information Systems
Business process changea study of methodologies, techniques, and tools
MIS quarterly
Productivity and performance effects of business process reengineeringa firm-level analysis
Journal of Management Information Systems
Potential pitfalls of process modelingpart a
Business Process Management Journal
Learning to specify information requirementsthe relationship between application and methodology
Journal of Management Information Systems
Cognitive fit in requirements modelinga study of object and process methodologies
Journal of Management Information Systems
Managing information about processes
Journal of Management Information Systems
Business process model abstraction : a definition, catalog, and survey
Distributed and Parallel Databases
Workflow view driven cross-organizational interoperability in a web service environment
Information Technology and Management
Fundamentals of Business Process Management
Business Process EngineeringReference Models for Industrial Enterprises
Metrics for Process ModelsEmpirical Foundations of Verification, Error Prediction, and Guidelines for Correctness
YAWLyet another workflow language
Information Systems
The Seas of Language
Theories of meaning in schema matchinga review
Journal of Database Management (JDM)
On the deep structure of information systems
Information Systems Journal
On sense and reference
Philosophical Review
On denoting
Mind
Cited by (24)
Guided Process Discovery – A pattern-based approach
2018, Information SystemsCitation Excerpt :It can be challenging to automatically assign good labels to discovered patterns. Methods for automatic labeling of process fragments [15] do exist. Work on decomposed process discovery [11,12] could be leveraged to obtain activity patterns that represent parts of the observed behavior.
HESML: A scalable ontology-based semantic similarity measures library with a set of reproducible experiments and a replication dataset
2017, Information SystemsCitation Excerpt :Mendling et al. [80] study the current practices in the activity labeling of business processes, whilst Dijkman et al. [32] propose a similarity metric between business process models based on an ad-hoc semantic similarity metric between words in the node labels and attributes, as well as the structural similarity encoded by the concept map topology. Likewise, Leopold et al. [68] propose an automatic refactoring method of activity labels in business process modeling based on the automatic recognition of labeling styles, and Leopold et al. [67] propose the inference of suitable names for business process models automatically. Finally, Montani and Leonardi [89] introduce a framework for the retrieval and clustering of process models based on a semantic and structural distance between models.
An experiment on an ontology-based support approach for process modeling
2017, Information and Software TechnologyCitation Excerpt :One of the major aids to support BPM are business process models. Many organizations have a dedicated BPM team, a competence center or a consulting department with BPM experts who are in charge of process modeling [3]. However, there are also many casual modelers involved with supporting BPM initiatives who are not extensively trained [4,5].
Business Process Retrieval from Large Model Repositories for Industry 4.0
2024, IEEE Transactions on Services ComputingAdvancing Business Process Science via the Co-evolution of Substantive and Methodological Knowledge
2022, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)Process model index construction and retrieval supporting structure and behavior fusion
2021, Jisuanji Jicheng Zhizao Xitong/Computer Integrated Manufacturing Systems, CIMS