Doctoral defence: Katsiaryna Lashkevich "Data-driven analysis and optimization of waiting times in business processes“

On 10 January at 10:15 Katsiaryna Lashkevich will defend her thesis "Data-driven analysis and optimization of waiting times in business processes“ to obtain the degree of Doctor of Philosophy (in Computer Science). 

Supervisors
Assoc. Prof Fredrik Payman Milani, University of Tartu 
Prof. Marlon Dumas, University of Tartu 

Opponents
Prof. Luise Pufahl, Technical University of Munich (Germany)
Prof. Remco Dijkman, Eindhoven University of Technology (Netherlands)

Summary
Waiting times are inevitable in business processes but, if ignored, can lead to significant inefficiencies. Consider your research paper, completed and ready to reach its audience but stalled on a reviewer’s desk — why? What causes waiting, and how can it be reduced? Most business processes are supported by software applications like manuscript handling systems used by reviewers. These systems track the activities of process participants, generating process execution data stored as event logs. Process mining techniques enable the analysis of such event logs, providing insights into process performance. 

This thesis proposes a set of process mining-based approaches to identify waiting time causes from event logs and recommend effective process redesigns. The first contribution is an approach to discover waiting times caused by batching (such as when papers are batched before review) and analyze them to reveal potential improvement opportunities. Further, the second contribution extends the analysis to other causes and introduces an approach to discover five causes of waiting times: batching, resource contention (reviewers handle other papers), prioritization (reviewers prioritize certain papers ahead of yours), resource unavailability (reviewers on vacation), and extraneous factors (reviewers handle tasks in other processes, e.g., prepare grant applications). Analysis of these causes can reveal where improvements shall be targeted. The third contribution is a method for fine-tuning a large language model to analyze the discovered causes of waiting times and recommend process redesigns (such as reducing the number of papers in a batch or balancing reviewer workloads more evenly) targeting identified causes. 

The proposed contributions are implemented in a software tool Kronos, enabling analysts to diagnose waiting time causes and obtain suggestions for process redesign. This is a step toward streamlining business processes, including getting your research published faster. 

The defence will be held also in Zoom (meeting ID: 988 4455 3111, passcode: ati)