Nautilus - The Process of Process Modeling

While process modeling has gained increasing importance for documenting business operations and automating workflow execution, process models display a wide range of quality problems impeding their comprehensibility and consequently hampering their maintainability. Literature reports, for example, error rates between 10% and 20% in industrial process model collections. Moreover, non-intention-revealing or inconsistent naming, redundant process fragments, and overly large and unnecessarily complex process models are typical quality problems which can be observed in existing process model collections. These problems have resulted in vivid research on the quality of process models with the goal of obtaining a better understanding of factors influencing the quality of process models. While existing research mostly focuses on the product or outcome of process modeling, the Nautilus project aims at taking a closer look on how process models are created (i.e., the process of process modeling). Obviously, factors influencing this modeling process eventually have an impact on the quality of its outcome (i.e., the resulting process model) and the incurred cost of its creation. The major goal of the Nautilus project is to systematically investigate the process of process modeling with the ultimate goal of improving the product of process modeling. To achieve our research objectives we will rely on a novel method for capturing and analyzing the process of process modeling. In particular, the Nautilus project will systematically investigate which modeling strategies are applied by both novices and experienced process modelers, trace these strategies back to process model quality, and analyze how novices and experienced process modelers differ in this respect. Moreover, Nautilus will investigate the impact of tool support (i.e., provision of high-level change patterns, refactoring support and automatic layout support) on the process of process modeling. The results of this investigation will then be used to develop novel methods and tools to support process modelers in creating process models through recommendations and consequently improving the product of process modeling. Moreover, the insights into the process of process modeling we will obtain from Nautilus are expected to provide important benefits for teaching.

While process modeling has gained increasing importance for documenting business operations and automating workflow execution, process models display a wide range of quality problems impeding their comprehensibility and consequently hampering their maintainability. Literature reports, for example, error rates between 10% and 20% in industrial process model collections. Moreover, non-intention-revealing or inconsistent naming, redundant process fragments, and overly large and unnecessarily complex process models are typical quality problems which can be observed in existing process model collections. These problems have resulted in vivid research on the quality of process models with the goal of obtaining a better understanding of factors influencing the quality of process models. While existing research mostly focuses on the product or outcome of process modeling, the Nautilus project aimed at taking a closer look on how process models are created (i.e., the process of process modeling, referred to as PPM). The central hypothesis of the Nautilus project hence can be formulated as follows: Factors that influence the creation of the PPM finally also influence the quality of the resulting process model and the incurred cost of its creation.The primary goal of the Nautilus project was to systematically investigate the process of process modeling with the ultimate goal of improving the product of process modeling, i.e., the process model. By developing and applying a novel method for capturing and analyzing the process of process modeling, the Nautilus project could identify reoccurring pattern of behavior to characterize the process of process modeling as well as distinct modeling styles and establish a link between certain modeling behavior and an improved modeling outcome. In other words, in the Nautilus project, it was possible to empirically establish that by following a certain strategy for creating a process model, higher quality of the resulting process model can be achieved. In this sense, the central hypothesis of the Nautilus project the connection between the PPM and the resulting process model quality could be empirically corroborated. Beside this scientific contribution, this finding also affects the education of future process modelers: instead of only teaching the meaning and usage of modeling elements, future process modelers can also be taught specific ways of creating a process model in order to achieve high-quality process models. Furthermore, in the Nautilus project, the role of software for developing process models, referred to as modeling tools, was investigated in detail. In these studies, it could be shown that there is no silver bullet for process modeling, as often claimed by tool vendors. Rather, depending on the specific modeling purpose, a specific approach should be followed. Besides the scientific relevance, these finding also impacts industry: future developments of tool vendors can be based on the scientific results obtained in the Nautilus project.