Summary (797511), страница 2
Текст из файла (страница 2)
Moscow: Institutefor System Programming RAS, 2014. P. 88–95. (Conference Paper; authorcontribution: 0,7) [15]Other Papers of the AuthorThe set of this thesis author’s publications is not limited to the paperslisted above. The following papers are related to the subject of process miningand analysis, but contain no of thesis’ contributions. As previously, papers aregrouped into three categories.First-tier Publications1. Rubin V., Mitsyuk A. A., Lomazova I. A., van der Aalst W. M.
P. ProcessMining Can Be Applied to Software Too!, in: Proceedings of the 8thACM/IEEE International Symposium on Empirical Software Engineeringand Measurement. NY: ACM, 2014. Ch. 57. P. 1–8. [16]7Second-tier Publications1. Nesterov R. A., Mitsyuk A. A., Lomazova I. A. Simulating Behavior ofMulti-Agent Systems with Acyclic Interactions of Agents //Proceedings of the Institute for System Programming.
2018. Vol. 30. No. 3.P. 285–302. [17]2. Shugurov I. S., Mitsyuk A. A. Applying MapReduce to ConformanceChecking // Proceedings of the Institute for System Programming. 2016.Vol. 28. No. 3. P. 103–122. [18]3. Nikitina N., Mitsyuk A. A. Carassius: A Simple Process ModelEditor // Proceedings of the Institute for System Programming. 2015. Vol.27. No. 3. P.
219–236. [19]4. Mitsyuk A. A., Kalenkova A. A., Shershakov S. A., van der Aalst W. M. P.Using process mining for the analysis of an e-trade system: Acase study // Business Informatics. 2014. Vol. 29. No. 3. P. 15–27. [20]Third-tier Publications1. Mitsyuk A. A., Kotylev Y. V. Layered Layouts for SoftwareSystems Visualization Using Nested Petri Nets, in: Tools andMethods of Program Analysis: 4th International Conference, TMPA 2017,Moscow, Russia, March 3-4, 2017, Revised Selected Papers. Communicationsin Computer and Information Science Vol.
779. Springer InternationalPublishing, 2018. Ch. 11. P. 127–138. [21]Conferences and WorkshopsThe results of this thesis have been presented and discussed at thefollowing conferences, seminars, and workshops:1. 2017 Ivannikov ISPRAS Open Conference. Moscow, Russian Academy ofSciences, 30.11.2017. Talk: Non-Local Correction of Process Models UsingEvent Logs.2. Seminar of the Moscow ACM SIGMOD Chapter. Moscow, MSU Faculty ofComputational Mathematics and Cybernetics, 26.10.2017.
Talk: Корректировка моделей процессов по логам событий (Process Model Correctionbased on Event Logs).83. Seminar of PAIS Laboratory. Moscow, NRU HSE Faculty of ComputerScience, 09.10.2017. Talk:Использование журналов событий для локальной корректировки моделей процессов (Using Event Logs for LocalCorrection of Process Models).4. 6th International Conference - Analysis of Images, Social networks and Texts(AIST 2017).
Moscow, Polytechnic University, 27-29.07.2017. Poster: ProcessModel Repair by Detecting Unfitting Fragments.5. Seminar of PAIS Laboratory. Moscow, NRU HSE Faculty of ComputerScience, 17.10.2016. Talk: Модульное исправление моделей процессов(Modular Process Model Repair).6. Spring/Summer Young Researchers’ Colloquium on Software Engineering2015 (SYRCoSE 2015).
Samara, Povolzhskiy State University ofTelecommunications and Informatics, 28-30.05.2015. Talk: Iskra: A Tool forProcess Model Repair.7. Special Seminar «Процессно-ориентированные информационные системы» (“Process-Aware Information Systems”). Moscow Oblast’, Voronovo, 2829.11.2015. Talk: Об исправлении моделей процессов (On Process ModelRepair).8. Seminar of PAIS Laboratory. Moscow, NRU HSE Faculty of ComputerScience, 12.10.2015. Talk: Плохие и хорошие модели бизнес-процессов(Bad and Good Business Process Models).9.
Spring/Summer Young Researchers’ Colloquium on Software Engineering2014 (SYRCoSE 2014). St. Petersburg, Peter the Great St. PetersburgPolytechnic University, 29-31.05.2014. Talk: Generation of a Set of EventLogs with Noise.Thesis OutlineThe thesis consists of the four main chapters, the introduction, and theconcluding chapter.Introduction discusses the relevance of thesis subject, informally presentsits main problem, contains necessary formal sections, and shortly outlines thethesis content.The first chapter describes the background of this thesis.9Firstly, Petri nets are defined.
This is the language that is used in orderto model processes in this thesis. Petri net is a directed bipartite graph withtwo node types. Transitions model activities of a process, and places are used tospecify the order of these activities. In particular, it is possible to model all basicprocess patterns: sequence, choice, parallel execution, and loops. These nodes areconnected with arcs.The current state of a Petri net is denoted using a so-called marking.
Amarking is a distribution of tokens through net places. A transition of a Petri netmay fire if there are enough tokens in all of its input places. Firing of a transitionconsumes tokens from its input places and produces tokens to its output places.Thus, a transition firing changes the marking of the net.In this thesis, workflow nets are used. They form a special class of Petrinets.
A workflow net has distinguished initial (source) and final (sink) places. Theinitial marking of a workflow net is a single token in the source, whereas in thefinal marking all places are clear except the sink one. All possible runs in such anet start from the initial marking and end in the final marking. Figure 3 shows anexample of a workflow net. A black token in the source place denotes the initialmarking.ht8asourcet1bc1t2c2cft3t6edc3t5t4c4gsinkt7Figure 3: Simple workflow netSecondly, event logs are defined.
Let ⊆ be a set of process activities.They are used to label model transitions. We assume that an event in a log is aname of an activity, i.e. additional event attributes are not used. Thus, an eventrepresents an activity in a trace of an event log.We define a trace as a finite sequence of activities from , i.e. ∈ * .An event log (or log) is a finite multi-set of traces, i.e.
∈ ℬ(* ). For example,[︀]︀ = ⟨,,,,⟩3 , ⟨,,,,⟩5 , ⟨,,,,⟩ is an event log with the same activitiesas in the model shown in Figure 3.Thirdly, process mining and its three sub-fields are described. Processmining is a research area at the intersection of formal methods and data science [5].It provides methods for analysis of processes in information systems based on eventlogs, which are recorded during their lifetime. The three main sub-fields of process10mining are process discovery, conformance checking, and process enhancement.Algorithms from each of these sub-fields are employed in this thesis.Process discovery algorithms are able to synthesize a process modelautomatically from a given event log. Conformance checking aims at evaluating themodel’s conformance to the observed behaviour that is recorded in an event log.Finally, both processes and process models can be enhanced.
Process enhancementprovides methods to improve real-life processes according to specified performancecriteria. Usually, these methods are less formal, and business-oriented. Processmodel enhancement methods and techniques to improve models are more formal.The process model repair (or simply model repair ) belongs to this category ofprocess mining methods [5].The four main numerical criteria of a process model are described in thischapter: fitness, precision, generalization, simplicity.
These criteria are commonin process mining [5].The first two are used for checking the conformance between a model andan event log. Fitness shows to what extent log traces can be replayed by a model.Precision determines how well the model can generate behaviour that is notrepresented in the log. Generalization and simplicity characterize the propertiesof the model itself. Generalization shows the level of model abstraction, andsimplicity shows its compactness.This thesis considers the problem of process model repair according to itsfitness. Precision criterion is also employed, when repaired models are evaluated.In this chapter, concrete methods for fitness and precision measurement aredescribed in detail.
In this thesis, an alignment-based technique to check logmodel conformance [22] is used, since this technique is the most conventional inprocess mining at the moment.This chapter also describes the different process discovery algorithms [5].Two of them are selected to be used in this thesis: Inductive miner [23] andILP miner [24].
The main reason to use these algorithms with carefully specifiedsettings is that both of them can guarantee perfect fitness of a discovered model.Fourthly, this chapter also reviews the model repair methods proposedearlier by other researches. Particular problem statements, solutions and theirfeatures are considered.The approach proposed in this thesis has its specific features. The closestapproach has been proposed by D. Fahland and W. van der Aalst [25]. However,we employ model decomposition techniques for the subsequent replacement ofnon-conforming model fragments. Thus, a repaired model differs from the originalonly by these replaced (repaired) fragments.
If discrepancies between an event logand a process model are local, then the change will be insignificant.11It is especially important to preserve a model structure in the cases whenthe original model has been built by an expert, and therefore this model is wellperceived by a human being.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
