Generalized Alignment-Based Trace Clustering of Process Behavior

Abstract

Process mining techniques use event logs containing real process executions in order to mine, align and extend process models. The partition of an event log into trace variants facilitates the understanding and analysis of traces, so it is a common pre-processing in process mining environments. Trace clustering automates this partition; traditionally it has been applied without taking into consideration the availability of a process model. In this paper we extend our previous work on process model based trace clustering, by allowing cluster centroids to have a complex structure, that can range from a partial order, down to a subnet of the initial process model. This way, the new clustering framework presented in this paper is able to cluster together traces that are distant only due to concurrency or loop constructs in process models. We show the complexity analysis of the different instantiations of the trace clustering framework, and have implemented it in a prototype tool that has been tested on different datasets.

Improving the performance of process discovery algorithms by instance selection

“…For computing the quality of discovered process models, the original event logs were used. We sampled event logs with different variant and trace-based instance selection strategies, and c with different values that are [1,2,3,5,10,15,20,25,30,40,50,75, 100] which c presents how many percentages of traces or variants of the event log is selected to be placed in the sampled event log. Therefore, if we use the trace-based sampling policy with c = 100, the sampled event log equates to the original event log.…”

“…We used sampled event logs just for discovery purpose, and the original event logs were used for computing F-Measure values. For the cases that the instance selection methods were used, we applied the sampled size in [1,2,3,5,10], and the average of F-Measure values is shown. For the statistical sampling method, we iterate the experiment four times, and again the average of F-Measure values are considered.…”

“…To deal with the complexity of process models, in some research, such as [42,31,51], clustering methods have been applied to get several sub-models according to clustered sub-logs. For example, [30] and [10] use data attributes and conformance artifacts for this purpose. Using this approach, the quality of the sub-models is better than a single process model discovered on the complete event log.…”

Encoding Conformance Checking Artefacts in SAT

“…The minimal distance between each trace and this full run is 9. We compared our result with the module Anti-Alignment of ProM 6 , that computes anti-alignment over Hamming distance. For the same size of run, the algorithm returned the sequence s, b, d, d, d, c, a, d that is indeed linearly far from the log traces.…”

“…Recent studies focus on SAT implementation of Data Mining algorithm in order to satisfy all the constrains and get optima [20,14]. By introducing a SAT implementation of alignments in this paper, we hope to push a new family of algorithmic methods for conformance checking in the line of [12,6]. However, these works mostly consider Hamming distance between log traces and process models, which is usually considered less appropriate than edit distance (c.f.…”