Facilitating Enhanced Decision Support Using a Social Norms Approach

Introduction

Automation in many domains based on business process models has progressed rapidly during the last few years. The rate of penetration differs from industry to industry and is still today led by manufacturing and supply chain management. The reason for a widespread penetration across diverse domains may not just be the maturity (De Bruin & Rosemann, 2005) of the respective industry but the reason also lies in the type of business processes (Inada, 1969). Highly standardized processes and well-defined supportive processes are conducive for automation. However, there remain the vast majority of processes that still need a knowledge worker in the loop to make decisions. The support that can be provided to these knowledge workers (Grünert et al., 2014) in the respective context using a social norms approach is the topic of this paper (Gomes et al., 2011).

Norms are expectations of behavior observed in human societies (Ullmann-Margalit, 1977). Norms provide guidance about how one has to behave in a certain situation (e.g. tipping in restaurants). However, there is also the possibility of violation. This is the freedom enjoyed by the individuals through their autonomy. There are different types of norms followed by societies such as prohibitions, obligations and permissions and these are often called as deontic norms (Meyer & Wieringa, 1993).

In this paper, we are interested in prohibition and obligation norms because these norms specify which actions are prohibited and obliged in a society (e.g. not littering a park and the obligation to buy a ticket when using the public transport). Prior research in human societies (Inada, 1969) shows that there are more prescriptive norms (i.e. obligations) than proscriptive norms (prohibitions). However, proscriptive norms were found to be stronger than prescriptive norms.

This paper takes a social BPM approach, where obligation and prohibition norms are identified in the context of decision making. The norms identified can then be applied to decision making of individuals when enacting a business process. Using this social approach, individuals are not confined to one choice but a range of choices (with an associated probability for the success of a chosen choice for a given problem at hand) and an informed decision can be made. For example, the user might be told that executing action B after A produces faults 90% of the time (i.e. prohibition with a probability of 90%) and that the choices that other users have made to substitute B for include D, E or F as these have a lower error rate associated with them. Additionally, the user is allowed to make an informed choice (i.e. they can choose action B if they think it is the right action to execute under prevailing conditions) which retains their autonomy and respects their decision-making capabilities. Thus, norms allow flexibility to human decision makers and we believe this is crucial in domains involving rapid changes in the functioning environment. Also, when situation change, the new norms can be identified and proposed to the human users which makes our approach both dynamic and flexible.

In order to motivate how our approach might be beneficial, let us consider the emergency prioritization system in hospitals. Patients arriving in the emergency department (ED) are assessed by a knowledge worker (e.g. a triage nurse) and are assigned different priorities based on the assessment. Assessments of individual cases can be complex and often decision support systems are used for prioritization. Our approach will enhance the existing system through a) learning from history of cases available in the system by looking particularly at failed instances and b) recommending the most suitable option (path) for treating a patient in question.