An Inferential Knowledge Model for the Digitalisation and Automation of Business Process Analysis

Abstract

In the course of digitalisation, the service sector is undergoing rapid  change, which is partly questioning the existing business models and  influencing the conditions of competition. The consulting industry, too, is  confronted with corresponding challenges despite the positive sales trend. The  use of advanced technology-based tools based on artificial intelligence or  analytical approaches could sustainably improve the competitive situation of  consulting providers. Consulting projects often require a comprehensive  analysis of business processes. The knowledge-based system iKnow presented  here has mechanisms of automated inference as well as machine learning  approaches. In contrast to process mining, it can also be used to support and  partially automate process analysis if no evaluable log data is available. In this  article, the previous concept of an inference-capable business process analysis  tool is further developed, its usefulness is demonstrated and evaluated using an  example process. With the help of a modeled knowledge base for process  analysis, BPMN models can be automatically examined for weak points on the  basis of analysis criteria and suitable improvement measures can be determined. Machine learning approaches can be used to continuously improve the analysis  results.

Research Motivation in the Context of Digital Consulting Technologies

Digitalisation can be described as a change from real processes and services to  information system based services [1]. Today it forms the basis of many innovative  business models which aim at the flexibilisation, individualisation and autonomisation  of services [2–4]. Bughin et al. [5] defines digitalisation as the ability to network people, devices and physical objects independently of time and place. The improved processing and analysis capability of information from large data sets leads to a  higher level of automation of processes and decisions.  

Consultancies are faced with the challenge of reacting to digital changes and using  new technologies to innovatively transform classic consulting processes and business  models and to adapt to the changed consulting market [6]. One possible approach is  the virtualisation of consulting processes. This means the personal on-site contact  between consultant and client is reduced to a reasonable minimum through the use of  technology [7]. The extreme case of a purely technology-based consulting approach  can be described as full automation [4, 7]. This requires suitable IT artefacts. The  generic tool iKnow presented in this article is based on a formalised knowledge  model, mechanisms of automatic inference and machine learning approaches [8]. Due  to the knowledge-intensive character of the consulting industry, iKnow can be used  specifically for knowledge processing and decision support within consulting  projects. In a first project, the analysis of business processes will be considered [9].  

In contrast to process mining, it is not assumed that the business processes under  consideration are IT-supported and that the log data can be evaluated for this purpose.  The results of process mining include the reconstruction and analysis of processes in  order to determine implicit and hidden process information from data of IT systems  [10]. In contrast to business process simulation, no simulation-relevant process data is  required. The simulation usually supports the assessment of dynamic process behavior  and achieves quantitative time- and cost-related result data, which concern the  improvement of the interaction between human-task-technology [11]. Neither  simulation nor process mining consider soft and qualitative influencing factors such  as implicit process knowledge of the process participants. Without supplementary  procedures, they also do not allow automatic interpretation of the results obtained or  decisions on process improvements [9]. In contrast, iKnow should serve as the basis  for a knowledge-based and inferentially capable business process analysis tool and  enrich the automated process analysis with context-related information and expert  knowledge on the respective process.

In this article, the previous concept is further developed and the evaluation of  iKnow for automated business process analysis in the context of consulting tasks is  presented. On the basis of defined analysis criteria, Business Process Model and  Notation (BPMN) models can be examined for weak points and then suitable  improvement measures can be determined automatically. With the help of iKnow’s automated learning approaches, context-related process information and expert  knowledge can be integrated so that the automated business process analysis can be  continuously improved over time.

Methodically, the processing of the contribution can be assigned to design oriented  business informatics. The design science research approach according to Peffers et al.  [12] is chosen as the framework. After the description of the research motivation, the  basics of iKnow as well as core elements of the existing concept of the business  process analysis tool are presented. The formal model in iKnow is adapted and extended in chapter 3 according to the requirements of an automated business process  analysis. The methodological development of the artifact is based on a formal-deductive analysis [13]. Considering the limited space, the main focus of this paper  will be on the conceptual design and in particular on the demonstration of the  feasibility and usefulness of the artifact for automated business process analysis. Finally, a prototype evaluates the designed tool using an example of a business  process. The prototype itself and the evolutionary procedure for developing the  prototype application are not described. The last chapter sums up the findings gained  and points out subsequent research work.