1 dag geleden - Universiteit Utrecht (UU) - Utrecht
Utrecht University's Faculty of Humanities is looking for a Postdoc position in the Horizon2020 project “Integrity” (0.8 FTE). Are you interested? Then please …
In the context of the EU H2020 project BPR4GDPR (Business Process Re-engineering and functional toolkit for GDPR compliance), a PhD position is open at the Analytics for Information System (AIS) group (www.win.tue.nl/ais/) in TU/eÂ¿s Department for Mathematics and Computer Science in the domain of Stream Process Mining.
The broader scope of the BPR4GDPR projectIn the last two decades the focus on process-orientation (e.g., process-aware information systems or BPM systems) has increased, while, with the incredible growth of event data (cf. 'Big Data'), it has become possible to use process mining, i.e., a posteriori analysis technique exploiting the information recorded in event logs, to discover models and check the conformance of existing ones. Indeed, most organisations have very limited knowledge about the reality happening throughout their day-to-day operation; process mining focuses on this kind of problem, with a view to assessing the organisational reality and reduce the gap between what is supposed to happen and what actually happens. The key facets of process mining are discovery, monitoring and improvement of real processes by extracting knowledge from the organisation's available data. Previous research has pointed large discrepancies between the idealized model and the process in reality. Moreover, process mining has shown that different models are possible for different and particular views on the process at hand.
The goal of BPR4GDPR is to support the implementation of a Privacy-Aware Process Mining Framework, seeking to meet requirements related to: transparency, being able to discover and integrate interpretable business procedures into a process model, i.e., to generate process models reflecting, as precisely as possible, an organisation's current modus operandi; compliance, automatically identifying 'business rules' for different perspectives; and accountability, spotting non-conformant executions. While checking the conformance between a process model and events in reality, two main concepts should be considered: real-time data and concept drift.
Additionally, advanced stream mining techniques will be implemented within the framework of this project to support real-time, single-passage mining of sensitive data such that businesses can fulfill the new requirements of GDPR (General Data Protection Regulation).Privacy-Aware Stream MiningFor the position, the PhD candidate is expected to work on stream process and data mining and will be supervised by dr.ing. Marwan Hassani and dr.ir. Boudewijn van Dongen. In the domains of today's evolving, IoT-based organisations, business rules (forming the de jure model) can easily get less fulfilled over the time. This is observed through deviations of the real process (reflecting the de facto model) as collected from the event data from the de jure model. Such deviations can be either intended or accidental. In both cases, one is interested in a real-time detection of these concept drifts and in an online reporting of their severity. In a similar setting, new requirements may arise, such that the process model discovered from the underlying event data will be outdated and in need to be continuously updated/adapted.
Furthermore, to respect the requirements of EU's GDPR (put into effect from the end of May 2018), businesses are not allowed to store user sensitive data unless clearly authorized by end users. Even in cases when an authorization is obtained, users will always keep the right of their profile data 'to be forgotten'. PositionIn the light of the above description, both (i) stream process mining and (ii) stream data mining solutions should be provided during the period of this position. The former should define concept drift detection and stream process mining novel techniques that update the statistics about the data-related behaviours. They should establish patterns to efficiently integrate recent information to improve the model with continuous update of discovered models and to report concept drifts from a given reference model or from the previously discovered models. The latter should advance efficient, single-passage privacy-aware stream data mining approaches that are able to extract useful insights of the data on the fly without compromising any privacy regulations.
The Analytics for Information Systems (AIS) group provides its long running expertise and experience across all challenges of BPM, stream data mining and process mining. In addition, BPR4GDPR the project has one more PhD position on model adaptability (that will closely collaborate on the project and solve the upcoming challenges jointly from different angles.
We are looking for a candidate that meets the following requirements: