Indulpet Miner: Combining Discovery Algorithms

Leemans, Sander J. J.; Tax, Niek; Hofstede, Arthur H. M. ter

doi:10.1007/978-3-030-02610-3_6

Cited by 11 publications

(10 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Split Miner [11] is the best performing process discovery technique on two of the four logs when we learn the probability distribution per marking from the training data, with on the other logs the Heuristics Miner [78] and the Inductive Miner with 20% filtering [53] being the best approach. An interesting observation can be made about the Indulpet Miner [54], which does not perform well on average performs but has a very large 95%-CI for all three logs, indicating that for some of the random train/test-splits the method generates quite accurate predictions but for others very inaccurate ones.…”

Section: Resultsmentioning

confidence: 98%

An interdisciplinary comparison of sequence modeling methods for next-element prediction

2020

Self Cite

View full text Add to dashboard Cite

Data of sequential nature arise in many application domains in forms of, e.g. textual data, DNA sequences, and software execution traces. Different research disciplines have developed methods to learn sequence models from such datasets: (i) in the machine learning field methods such as (hidden) Markov models and recurrent neural networks have been developed and successfully applied to a wide-range of tasks, (ii) in process mining process discovery techniques aim to generate human-interpretable descriptive models, and (iii) in the grammar inference field the focus is on finding descriptive models in the form of formal grammars. Despite their different focuses, these fields share a common goal -learning a model that accurately describes the behavior in the underlying data. Those sequence models are generative, i.e, they can predict what elements are likely to occur after a given unfinished sequence. So far, these fields have developed mainly in isolation from each other and no comparison exists. This paper presents an interdisciplinary experimental evaluation that compares sequence modeling techniques on the task of nextelement prediction on four real-life sequence datasets. The results indicate that machine learning techniques that generally have no aim at interpretability in terms of accuracy outperform techniques from the process mining and grammar inference fields that aim to yield interpretable models.

show abstract

Section: Resultsmentioning

confidence: 98%

An interdisciplinary comparison of sequence modeling methods for next-element prediction

2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…Based on this framework, many variants (i.e. discovery techniques) have been proposed to handle various types of event logs and discovery challenges, such as [12,14]. The framework is robust and provides several formal guarantees, such as soundness (i.e.…”

Section: Related Workmentioning

confidence: 99%

“…Other process discovery techniques have been proposed, but these are either not applicable directly to abstractions (e.g. Evolutionary Tree Miner [3], Split Miner [2], Indulpet Miner [14]) or do not provide basic guarantees such as soundness (e.g. Split Miner [2], Integer Linear Programming Miner [22], Heuristics Miner [28]).…”

Section: Related Workmentioning

confidence: 99%

“…Indulpet Miner (IN) [14] (our experiment extends the evaluation in [14]), (3) Evolutionary Tree Miner (ETM) [3], (4) Split Miner (SM) [2], and (5) the baseline flower model (F) that allows for any behaviour. Therefore, we apply 7 process discovery techniques: for each log, we perform 3-fold cross validation: of these three parts, two parts are used for discovery while the remaining part is used for evaluation (measuring fitness [20], ETC-precision [15], simplicity).…”

Section: Model Qualitymentioning

confidence: 99%

See 1 more Smart Citation

Process discovery using in-database minimum self distance abstractions

Syamsiyah

Leemans

2020

Proceedings of the 35th Annual ACM Symposium on Applied Computing

Self Cite

View full text Add to dashboard Cite

Process executions generate event data that are typically stored in legacy information systems, such as databases. However, process discovery, which requires such event data, is performed in main memory. To bridge this gap, existing techniques must transform and extract event data, which can be expensive steps. This issue has been addressed by processing the event data directly in their origin. However, existing methods rely only on the simplest event data abstraction: the Directly Follows (DF) abstraction. This paper improves upon these existing works by considering another abstraction, the Minimum Self Distance (MSD) abstraction, which enables discovery of a larger class of models than the DF alone. That is, we propose IMw, a process discovery technique without logs and uses both the MSD and DF abstractions. Furthermore, this work proposes an approach to compute the MSD abstraction in-database, thus avoiding the need for transforming and moving event data. We evaluate IMw with real-life logs, and the experimental results show that IMw with in-database abstraction is faster than the traditional approach, aware of dynamic updates on event data, and able to discover models with pareto-optimal results, compared to existing techniques.

show abstract

“…In this paper, we present an approach to simplify process models, in particular abstract hierarchical representations of block-structured workflow nets (process trees) [19]. Such models result from process discovery techniques [4,18,20] or can be derived from Petri nets [25,31]. That is, we present a set of reduction rules that reduce the structural complexity of process models while keeping the visible behaviour of the models, that is, their language, the same.…”

Section: Introductionmentioning

confidence: 99%

Language-Preserving Reduction Rules for Block-Structured Workflow Nets

Leemans¹

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Process models are used by human analysts to model and analyse behaviour, and by machines to verify properties such as soundness, liveness or other reachability properties, and to compare their expressed behaviour with recorded behaviour within business processes of organisations. For both human and machine use, small models are preferable over large and complex models: for ease of human understanding and to reduce the time spent by machines in state space explorations. Reduction rules that preserve the behaviour of models have been defined for Petri nets, however in this paper we show that a subclass of Petri nets returned by process discovery techniques, that is, block-structured workflow nets, can be further reduced by considering their block structure in process trees. We revisit an existing set of reduction rules for process trees and show that the rules are correct, terminating, confluent and complete, and for which classes of process trees they are and are not complete. In a real-life experiment, we show that these rules can reduce process models discovered from real-life event logs further compared with rules that consider only Petri net structures.

show abstract

Indulpet Miner: Combining Discovery Algorithms

Cited by 11 publications

References 28 publications

An interdisciplinary comparison of sequence modeling methods for next-element prediction

An interdisciplinary comparison of sequence modeling methods for next-element prediction

Process discovery using in-database minimum self distance abstractions

Language-Preserving Reduction Rules for Block-Structured Workflow Nets

Contact Info

Product

Resources

About