Process-Awareness of Event-Driven Systems: Model, Analysis and Optimisation
Software Engineering and Programming Languages
Final Report Abstract
In domains such as healthcare, logistics, or commerce, information systems support processes, loosely defined as the coordinated execution of a set of actions to achieve a certain goal. Traditionally, such systems leverage process models to orchestrate the execution of the actions of a process. While such an approach works well for relatively static and closed application scenarios, it is limited in its support for processes that shall be guided dynamically, in response to data that is sensed from diverse sources. In such scenarios, techniques for event stream processing (ESP) may help to achieve effective support of processes. By evaluating queries over streams of event data, ESP enables continuous monitoring of the progress of process execution and, based thereon, provides means for reactive control. Using ESP in process-oriented applications, however, raises several questions concerning the relation between essential concepts of processes and event streams, as well as linked to models that may enable the analysis and optimisation of the respective ESP systems. The project “Process-Awareness of Event-Driven Systems: Model, Analysis, and Optimisation”, PAEDS for short, set out to investigate these research questions. The main scientific results of the project are summarized as follows: In PAEDS, we phrased the abstraction of low-level events needed to identify the execution of actions of a process as the problem of discovering event queries from labelled data. Moreover, we developed the IL-Miner, an algorithm to address this problem. ◦ We presented interval-based models, coined temporal network representations, for event data of processes, which naturally capture the durations of actions along with their interplay. ◦ Formalising abstractions of event data as fragments of process models, we developed algorithms for the discovery of frequent fragments of process behaviour. ◦ To obtain models for quantitative analysis from low-level event data, we developed algorithms for property-preserving generalization of performance models. ◦ To achieve best-effort event stream processing in overload situations, we developed state-based load shedding strategies that exploit the regularities in event streams produced by processes. ◦ Targetting the efficient integration of data from remote sources in event stream processing, we devised algorithms for data prefetching and caching based on stream characteristics. The above results lay the foundations for using event stream processing in process-oriented applications. The feasibility of our techniques has been demonstrated in numerous experimental evaluations. They highlight the specific benefits in terms of enabling novel types of insights or improving the effectiveness and efficiency of existing analysis techniques. At a more abstract level, we draw two main conclusions from our experience in the PAEDS project. First, a main challenge when working at the intersection of process-oriented systems and event stream processing is the generalizability of research results. Despite well-defined notions and base concepts for processes and event streams, there is a wide variety of specific realizations of these notions in practice and, hence, also of their integration and interplay. As such, it is of utmost importance to clearly spell out the assumptions imposed by models, analysis techniques, optimization strategies on the relation between processes and event streams. Second, the potential for impactful research on process-awareness in event stream processing turned out to be even larger than expected. In all three areas explored in PAEDS, i.e., integrated modelling, quantitative analysis, and optimisation strategies, various novel research questions and ideas emerged. They provide a valuable starting point for follow-up work, which, for instance, more focus in more detail on distributed infrastructure, privacy considerations, and data quality issues.
Publications
- Il-miner: Instance-level discovery of complex event patterns. Proc. VLDB Endow., 10(1):25–36, 2016
Lars George, Bruno Cadonna, and Matthias Weidlich
(See online at https://doi.org/10.14778/3015270.3015273) - Integrating processoriented and event-based systems (dagstuhl seminar 16341). Dagstuhl Reports, 6(8):21–64, 2016
David M. Eyers, Avigdor Gal, Hans-Arno Jacobsen, and Matthias Weidlich
(See online at https://doi.org/10.4230/DagRep.6.8.21) - Pˆ3 -folder: Optimal model simplification for improving accuracy in process performance prediction. In Business Process Management - 14th International Conference, BPM 2016, Rio de Janeiro, Brazil, September 18-22, 2016. Proceedings, volume 9850 of Lecture Notes in Computer Science, pages 418–436. Springer, 2016
Arik Senderovich, Alexander Shleyfman, Matthias Weidlich, Avigdor Gal, and Avishai Mandelbaum
(See online at https://doi.org/10.1007/978-3-319-45348-4_24) - Complex event recognition languages: Tutorial. In Proceedings of the 11th ACM International Conference on Distributed and Event-based Systems, DEBS 2017, Barcelona, Spain, June 19-23, 2017, pages 7–10. ACM, 2017
Alexander Artikis, Alessandro Margara, Martin Ugarte, Stijn Vansummeren, and Matthias Weidlich
(See online at https://doi.org/10.1145/3093742.3095106) - Temporal network representation of event logs for improved performance modelling in business processes. In Business Process Management - 15th International Conference, BPM 2017, Barcelona, Spain, September 10-15, 2017, Proceedings, volume 10445 of Lecture Notes in Computer Science, pages 3–21. Springer, 2017
Arik Senderovich, Matthias Weidlich, and Avigdor Gal
(See online at https://doi.org/10.1007/978-3-319-65000-5_1) - Complex event processing under constrained resources by state-based load shedding. In 34th IEEE International Conference on Data Engineering, ICDE 2018, pages 1699–1703. IEEE Computer Society, 2018
Bo Zhao
(See online at https://doi.org/10.1109/ICDE.2018.00218) - Online temporal analysis of complex systems using IoT data sensing. In 34th IEEE International Conference on Data Engineering, ICDE 2018, Paris, France, April 16-19, 2018, pages 1727–1730. IEEE Computer Society, 2018
Avigdor Gal, Arik Senderovich, and Matthias Weidlich
(See online at https://doi.org/10.1109/ICDE.2018.00224) - To aggregate or to eliminate? Optimal model simplification for improved process performance prediction. Inf. Syst., 78:96–111, 2018
Arik Senderovich, Alexander Shleyfman, Matthias Weidlich, Avigdor Gal, and Avishai Mandelbaum
(See online at https://doi.org/10.1016/j.is.2018.04.003) - Context-aware temporal network representation of event logs: Model and methods for process performance analysis. Inf. Syst., 84:240–254, 2019
Arik Senderovich, Matthias Weidlich, and Avigdor Gal
(See online at https://doi.org/10.1016/j.is.2019.04.004) - Dynamic decision making for demand response through adaptive event stream monitoring. In 2019 IEEE Power & Energy Society General Meeting (PESGM), pages 1–5. IEEE, 2019
Gururaghav Raman, Jimmy Chih-Hsien Peng, Bo Zhao, and Matthias Weidlich
(See online at https://doi.org/10.1109/PESGM40551.2019.8974095) - Efficient discovery of compact maximal behavioral patterns from event logs. In Advanced Information Systems Engineering - 31st International Conference, CAiSE 2019, Rome, Italy, June 3-7, 2019, Proceedings, volume 11483 of Lecture Notes in Computer Science, pages 579–594. Springer, 2019
Mehdi Acheli, Daniela Grigori, and Matthias Weidlich
(See online at https://doi.org/10.1007/978-3-030-21290-2_36) - Introduction to the special issue on integrating process-oriented and event-based systems. Inf. Syst., 81:179–180, 2019
David M. Eyers, Avigdor Gal, Hans-Arno Jacobsen, and Matthias Weidlich
(See online at https://doi.org/10.1016/j.is.2019.01.004) - Load shedding for complex event processing: Input-based and state-based techniques. In 36th IEEE International Conference on Data Engineering, ICDE 2020, Dallas, TX, USA, April 20-24, 2020, pages 1093–1104. IEEE, 2020
Bo Zhao, Nguyen Quoc Viet Hung, and Matthias Weidlich
(See online at https://doi.org/10.1109/ICDE48307.2020.00099) - EIRES: Efficient integration of remote data in event stream processing. In SIGMOD ’21: International Conference on Management of Data, Virtual Event, China, June 20-25, 2021, pages 2128–2141. ACM, 2021
Bo Zhao, Han van der Aa, Thanh Tam Nguyen, Quoc Viet Hung Nguyen, and Matthias Weidlich
(See online at https://doi.org/10.1145/3448016.3457304)
