| dc.description.abstract | Municipalities deliver essential social services through complex processes that often suffer from inefficiencies and bottlenecks. Municipal information systems record detailed execution data on these processes. However, the data is often fragmented and noisy, making it difficult to analyze the data systematically and identify improvement opportunities.
In response to these challenges, this thesis presents a process mining-based framework for extracting process knowledge from raw municipal data in the social domain. The framework defines an end-to-end pipeline that preprocesses, transforms, and integrates fragmented municipal process data into structured XES event logs and regulatory patterns. From these, the framework systematically generates descriptive process models via process discovery techniques (reflecting actual execution behavior) and normative process models via the regulatory patterns (representing intended workflows).
By applying conformance checking and additional process diagnostics, the framework identifies deviations between actual and intended process behavior, as well as other actionable insights to potentially improve these processes. The framework is implemented and demonstrated on two municipal datasets, illustrating how the pipeline operates on real-world data and detailing the results
of each step of the framework.
To ensure reproducibility, the framework’s design and implementation are made publicly available, with each step documented in detail. The case study datasets and all resulting outputs of the framework, including the structured XES event logs, descriptive and normative models, and diagnostic results, are also published for transparency and future use. | |
