This is the homepage of the PINPOINT project.
PINPOINT is a project funded by MIUR under the PRIN 2020 initiative (grant B87G22000450001). The project aims at developing a full-fledged set of techniques towards explainable, knowledge-aware process intelligence. It does so by relying infusing process mining with recent advancements in artificial intelligence and declarative languages and techniques to represent flexible processes.
Brief Description of the Project
One of the most important goals of contemporary organizations, ranging from private companies to public institutions, is to continuously improve and optimize their operational processes, incrementing customer satisfaction and the quality of products/services, while reducing costs and increasing performance. This is traditionally tackled using techniques like Six Sigma or Total Quality Management, where business analysts only get an indirect reconstruction of the organizational reality through interviews, reports, on-field studies.
The massive digitalization of contemporary organizations and, more in general, of our societal ecosystem, is a game-changer in this spectrum: every performed activity leaves a digital footprint, which often provides a faithful representation of how processes are effectively executed in practice. Process mining is an innovative field at the intersection of model-driven engineering and data science, whose purpose is to analyse such event data to obtain insights on how processes are executed in reality, and trigger process improvement based on factual evidence.
Through process mining techniques, it is now possible to carry out sophisticated tasks such as:
- process discovery, automatically learning a representative process model from event data, reconstructing the actual process (not the desired process that management expects to be executed);
- conformance checking, to measure the deviation between the expected behaviours contained in a process model and the actual event data;
- predictive monitoring, to estimate how a running process will progress, based on partial observations on the current status.
The applicability of those techniques relies on the presence of high-quality, explicit event data that have to be harvested from the heterogeneous systems employed by the organisation, making the “garbage-in garbage-out” factor an actual risk: if event data are not of good quality, are largely incomplete, and contain missing or wrong information, then the results produced through process mining pipelines are misleading and unfaithful. At the same time, the employed process mining techniques contain black-box algorithms that do not provide explanations about the produced results, nor are able to properly incorporate organizational knowledge about the specific domain in which they are used.
The PINPOINT project tackles these limitations by developing a full-fledged set of techniques towards explainable, knowledge-aware process intelligence. It does so by relying on recent advancements in artificial intelligence and declarative languages and techniques to represent flexible processes.