Agent Assignment for Process Management: Agent Performance Evaluation Framework

Abstract

Convergence of data mining and process management is ideal -but still limited. An example of such a convergence is presented in the form of APE Framework that addresses the problem of static-agent-assignment-strategies in Workflow Management Systems (WfMS) -one cause of poor business process performance since all eligible agents may be assigned to a task instead of only assigning those which are expected to be "successful". To solve this problem, the APE Framework first identifies successful agents by investigating the history of workflow executions with methods from data mining and taking the process model into account as a source for domain knowledge. It then updates its findings into an organizational database -a data source wherefrom WfMS make assignments. Thus, through this convergence, the APE Framework enables WfMS to allocate only successful agentsinstead of merely static list of all eligible agents.

A Paradigm-Shifting From Domain-Driven Data Mining Frameworks to Process-Based Domain-Driven Data Mining-Actionable Knowledge Discovery Framework

“…EDM techniques [34] are not process-centric and don't concentrate on the event log generated data. On the other hand, EPM [35] is a process-centric approach, thereby considering the event log generated data to make the hidden patterns more erudite and actionable while taking decisions. These techniques help to analyze the students' performance and facilitate the directors, faculty instructors, advisors, and policymakers to improve the education process [36].…”

Human resource allocation in business process management and process mining

“…creation patterns, push and pull patterns), which can help demonstrate how resources can be used and represented in workflow systems. These resource patterns have been used across a variety of proposed approaches, providing support to the different allocation methods presented (Arias et al, 2015;Cabanillas et al, 2013Cabanillas et al, , 2015Stefansen et al, 2008;Talib et al, 2010;Tan and van der Aalst, 2006). For instance, Stefansen et al (2008) utilize distinct resource patterns as part of the resource allocation language called SOFTALLOC, in order to be able to manage different restrictions (soft constraints) during dynamic resource allocation.…”

“…Furthermore, other techniques have been used by distinct resource allocation methods, including Machine Learning algorithms (Huang, Lu and Duan, 2011; Huang, van der Aalst, Lu and Duan, 2011; Liu et al , 2008; Ly et al , 2005; Talib et al , 2010; Xu et al , 2014; Yingbo et al , 2007), Markov models (Huang, van der Aalst, Lu and Duan, 2011; Koschmider et al , 2011; van Hee et al , 2007), data mining techniques (Huang, Lu and Duan, 2011; Sindhgatta et al , 2016), constraints based-approaches (Barba et al , 2011, 2013), and multi-agent systems (Hsieh and Lin, 2014; Kress et al , 2007), among others.…”