Application of Process Mining in Logistic Processes of Manufacturing Organizations: A Systematic Review

Abstract

With the continuous development in technology and changes in the logistic systems, organizations should review their logistic processes that evolve over time. To attain a good insight into the conformance of these processes with the designed process model, constant detection and monitoring is required. The main objective of this systematic review is to investigate the state of the art in process mining applications in logistics specifically related to manufacturing organizations. The review aims to analyze and assess the use of process mining techniques and models in the logistics domain, based on the selected studies. In this review, literature was searched between the years 2004 and 2022 using several inclusion and exclusion criteria. Fifteen published studies were selected and analyzed on the use of process mining in the logistics domain based on the process mining techniques and models used. All of the selected studies used models and thirteen of them used real case studies. All of the fifteen studies used one or more algorithms. Only three of these studies did not mention the modeling language used to represent the process. Moreover, seven studies focused on discussing the process discovery alone and five more addressed process discovery in addition to other types of process mining. Eight studies mentioned the process mining tools used that included DISCO and several versions of ProM.

1. Introduction

Organizations strive to effectively and constantly improve the performance of various business processes by finding new techniques to manage and analyze the large amounts of event data available. Logistic processes are one of these complex and uncertain business processes involving the generation of massive amounts of event data, necessary for making accurate and timely decisions [1,2]. The complexity of these processes is related to the massive amount of goods and services handled and the significant number of people involved in the logistic system. With the ever-changing technologies and recurring innovations, business and logistic processes should cope rapidly with the new changes. However, with a highly dynamic logistic system, organizations face a plethora of challenges and deviations from planned processes that have to be managed effectively [3]. Analyzing these processes is the key for problem-solving and optimal performance [2]. Research reveals inconsistency between the line of action of logistic processes designed and the actual implementation of these processes. There is a substantial number of studies that regularly aims to analyze and develop logistic processes using different techniques, methodologies, and tools [4].

Process mining is one of the prominent research fields that are used to minimize errors caused by human bias and unrealistic judgments [5]. It allows acknowledging the real operational processes and supports their analysis [6]. Process mining is the nexus that lies between data mining and process analysis and monitoring. Starting with event logs, process mining uses real data to effectively and rapidly discover, track, and optimize business processes. Process mining involves three main types: discovery, conformance checking, and enhancement [5]. The process discovery type, which is the most distinguished process, involves analyzing the event log to create a more mature and updated process model that does not need any a priori information to be produced [5,7]. Since it cannot be predicted that all activities are present in an event log, process discovery is a challenging process [7].

Moreover, the conformance checking technique involves investigating the coherence of the event log with the existing process model to discover any deviations from reality. Enhancement is the last technique of process mining that involves the necessary improvements or extensions, using timestamps, that allow minimizing deviations such as bottlenecks [5,7]. Accordingly, process mining may be very beneficial when applied in the context of logistics of manufacturing organizations to frame process models. Through the techniques associated with process mining, organizations can discover whether a process model is as it is supposed to be as planned, check its conformance and alignment with reality, or enhance it by either modifying or extending the a priori model [6,8].

The main objective of this systematic review is to investigate the state of the art in process mining applications in logistics specifically related to manufacturing organizations. The review aims to analyze and assess the use of process mining techniques and models in the logistics domain, based on the selected studies. To the best of the researcher’s knowledge, there is no prior systematic review study has been conducted on the use of process mining in logistics; thus, indicating a gap in literature. Accordingly, this systematic review will contribute to the knowledge and fill the research gap on this issue.

This systematic literature review was guided by the guidelines of [9] in performing systematic literature reviews. Moreover, different elements of process mining such as the algorithms, modeling languages, types of process mining, and tools were used to analyze selected studies. The second section in this study illustrates the research method used to analyze the research studies and execute this systematic literature review. As for the third section, it includes the results and discussion of the searched studies, whereas the fourth section contains a conclusion.

2. Research Method

In this section, the methodology of this systematic literature review was utilized to exhibit the existing knowledge concerning how process mining and its elements are applied in logistics and what relevant knowledge is overlooked. This review was performed in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses, see Supplementary Materials) guidelines [10]. The study was registered in OSF; registration number https://0-doi-org.brum.beds.ac.uk/10.17605/OSF.IO/R3PKB, accessed on 6 July 2023. This study aimed to collect and analyze studies in this area of research. Guidelines from [9] were adopted in this systematic review. According to these guidelines, research questions were formulated and used to initiate the search phase of this process followed by database selection to automatically detect and find papers of research studies that are relevant to the search topic [9]. Moreover, a more refined search was conducted using the inclusion and exclusion criteria and primary studies were selected. This process was followed by data extraction from the set of selected studies that reveal the current state of the art research on process mining in the context of logistics. The final step was analyzing the data collected to distil results [9]. This process is discussed in more detail.

2.1. Research Questions

The research questions for this systematic literature review are based on extracting the knowledge provided in literature about the use of process mining and its techniques and elements in logistics of manufacturing organizations. Logistic processes are very complex and need to be analyzed and synthesized. There is no known systematic review that was conducted specifically on the use of process mining techniques in logistics of manufacturing organizations. The formulated research questions for this systematic review are as follows:

• RQ1. What is the knowledge provided in literature about the application of process mining in logistic processes?
• RQ2. What are the process mining techniques and algorithms used in logistics and available in the literature?

By answering these questions through this systematic review, it will become possible for researchers and decision-makers to find knowledge about process mining and its available techniques and algorithms used in the logistics of manufacturing organizations. In addition, this systematic review pinpoints gaps in current research about process mining in logistic processes, which can direct new investigation schemes [9].

2.2. Search String

The search approach for this systematic review was based on using an automatic search to explore a variety of databases in literature. The main key terms used in this search were: “process mining”, “logistics”, and “manufacturing”. However, the synonym “industry” was used for the term “manufacturing”. Using Boolean operators, the search key terms used were as follows: (“process mining” AND “logistics” AND “manufacturing”) OR (“process mining” AND “logistics” AND “Industry”). These two sets of key terms were used to find articles in a group of academic databases as directed in the guidelines of [9]. A set of five databases were selected and searched for published articles and conference papers and proceedings from the year 2004 until recent. English was chosen as the main language for this search before using the key terms in the searching process and duplicates were removed. The selection criteria of more than one academic database allowed attaining all the possible studies that were conducted on the topic discussed.

Table 1. Databases and initial search.

Database	Initial Automated Search Using (“Process Mining” AND “Logistics” AND “Manufacturing”)	Initial Automated Search Using (“Process Mining” AND “Logistics” AND “Industry”)	Total
IEEE Xplore	101	173	274
SpringerLink	7255	9624	16,874
ScienceDirect	5731	7791	13,522
Scopus	110	253	363
Google Scholar	48,900	81,500	130,400
Total Number of Initial Search	76,288	99,341	175,629

illustrates these databases and the initial automated search using the two groups of key terms.

2.3. Inclusion and Exclusion Criteria

The search process was continued by refining the selection using the inclusion and exclusion criteria based on [9] guidelines. As described in Table 1, the total number of papers in the initial search was 175,629 research papers, which is considered a huge number. A manual process was required to find and select the relevant studies and include them in this review and exclude irrelevant papers. The manual selection was grounded on the subsequent inclusion and exclusion criteria. According to the guidelines of [9], the inclusion criteria included papers that: 1. Investigate a methodological issue related to the topic of the systematic review, 2. Include instruments evaluating the quality of primary studies or strength of evidence, 3. Can make improvements in the field, and 4. Are in the English language. In this review, the included papers were chosen to be in the English language from the beginning and: 1. Are primary research studies offering unbiased evidence related to the research question, 2. Focus mainly on the use of process mining in logistics, and 3. Propose a model/algorithm/framework that relate to the application of process mining in logistics. The exclusion criteria related to guidelines in [9] included papers that: 1. Are secondary or tertiary studies presenting the outcomes of a systematic review or a mapping study, 2. Have a general focus, 3. Are represented by abstracts or presentations, 4. Provide instructions for conducting a primary study. The exclusion criteria allowed excluding papers that: 1. Are secondary research studies and reviews, 2. Discuss either process mining or logistics, 3. Are not conference papers or include case studies, and 4. Are providing instructions for performing primary studies. To achieve the inclusion and exclusion criteria, the following steps had to be followed: 1. Performing an extensive overview of the titles of the papers along with the abstracts to examine their relevance to the topic, and 2. If some of the studies are found to be irrelevant but need further inspection before excluding them, the full text is read and analyzed.

3. Results and Discussion

The results of the initial search revealed that the studies that included the key terms “process mining” AND “logistics” AND “manufacturing” were 76,288 (101 from IEEE Xplore, 7255 from SpringerLink, 5731 from ScienceDirect, 110 from Scopus, and 48,900 from Google Scholar). However, the key terms “process mining” AND “logistics” AND “Industry”, yielded 99,341 papers (173 from IEEE Xplore, 9624 from SpringerLink, 7791 from ScienceDirect, 253 from Scopus, and 81,500 from Google Scholar). All these papers were in the English language and were published starting from 2004 to 2022. After applying the inclusion and exclusion criteria, the included studies investigated process mining in logistic processes of manufacturing organizations, whether it is within or across organizations, and the elements of process mining utilized in every study. However, the studies that did not relate to the logistic processes in industry or manufacturing organizations, and did not include process mining techniques, or those related to the management of logistic processes and conducted in logistic companies were excluded.

As illustrated in Figure 1, many steps were taken for the selection of relevant papers that fulfill the needs of this systematic review. After removing irrelevant papers and duplicates, the remaining studies, which were reduced to 87 papers, were thoroughly reviewed using the contents of the full text. The final selection resulted in obtaining 15 papers that are fully relevant to the topic of this systematic review, which addressed the techniques used in logistics and related to process mining.

Furthermore, all the titles of the selected studies, their authors, date of publication, and type of publication were illustrated in

Table 2. Final selection of studies.

Ref.	Title	Authors	Date of Publication	Type of Publication
[1]	Performance of an Automated Process Model Discovery—the Logistics Process of a Manufacturing Company	Halaska and Sperka	2019	Journal
[11]	Enabling Value Stream Mapping for Internal Logistics Using Multidimensional Process Mining	Knoll et al.	2019a	Journal
[12]	Process Discovery Method in Dynamic Manufacturing and Logistics Environments	Intayoad et al.	2020	Conference paper
[4]	Applying Process Mining in Manufacturing and Logistic for Large Transaction Data	Intayoad and Becker	2018b	Conference paper
[13]	Evaluating the use of the open trip model for process mining: An informal conceptual mapping study in logistics	Piest et al.	2021	Conference-position paper
[14]	Optimization of Logistics Processes by Mining Business Transactions and Determining the Optimal Inventory Level	Terlouw	2017	Conference paper
[3]	Context Aware Process Mining in Logistics	Becker and Intayoad	2017	Conference paper
[15]	Developing an Internal Logistics Ontology for Process Mining	Knoll et al.	2019b	Conference paper
[16]	Trace Clustering Exploration for Detecting Sudden Drift: A Case Study in Logistic Process	Prathama et al.	2019	Conference paper
[17]	Mapping Log Data Activity Using Heuristic Miner Algorithm in Manufacture and Logistics Company	Pane et al.	2021	Journal
[6]	An Agent-Based Process Mining Architecture for Emergent Behavior Analysis	Bemthuis et al.	2019	Workshop
[8]	Process Mining in Logistics: the Need for Rule-Based Data Abstraction	van Cruchten and Weigand	2018	2018
[18]	Data-Based Description of Process Performance in End-To-End Order Processing	Schuh et al.	2020	Journal
[2]	Exploring the Relationship between Business Processes and Contextual Information in Manufacturing and Logistics Based on Event Logs	Intayoa andBecker	2018a	Conference paper
[19]	A Classification of Process Mining Bottleneck Analysis Techniques for Operational Support	Bemthuis et al.	2021	Conference-position paper

.

As described in Table 2, Wacharawan Intoyoad and Till Becker conducted three of all the selected studies and shared in one more. Dino Knoll and Gunther Reinhart shared in two studies, Rob Henk Bemthuis and Faiza Allah Bukhsh shared in three studies, and Jean Paul Sebastian Piest shared in two studies. However, Wil M.P. van der Aalst, the father of process mining, shared in only one study [18] about the application of process mining in logistics from the 15 studies analyzed in this paper.

Moreover, the number of conference papers selected is 10, representing about 67% of all the selected studies. Although the search included studies from 2004, all studies that were relevant to the topic of process mining and logistics of manufacturing companies began from the year 2017. As shown in Table 2, two studies of [4,13] were conducted in 2017, three studies of [2,8,16] were published in 2018, five studies of [1,3,11,14,17] were conducted in 2019, two studies of [6,18] were published in 2020, and three studies of [12,15,19] were published in 2021. Figure 2 is a chart representing the years searched and the number of publications in each of these years.

It is also notable to analyze the selected papers on the basis of the countries of their authors who have added to the base of knowledge in the field of process mining in logistics. Moreover, this analysis demonstrated the research groups interested in this area of research and their global distribution. As mentioned in

Table 3. Authors by country and selected papers.

Number of Authors by Country	Number of Selected Papers
11 from Germany	7
13 from The Netherlands	6
2 from Czech Republic	1
2 from South Korea	1
6 from Indonesia	2
1 from Sri Lanka	1
35	18

, the 18 selected papers were conducted by 35 authors from six countries. Some of these authors, as previously mentioned, have shared in more than one paper.

Concerning the topic of this systematic review, 37% of the authors were from The Netherlands, 31% from Germany, 17% from Indonesia, 6% from the Czech Republic and South Korea, and 3% from Sri Lanka.

Table 4. Analysis of selected studies on the basis of purpose, methodology, and outcomes.

Ref.	Authors and Year	Purpose and Methodology	Outcomes and Contribution
[1]	Halaska and Sperka, 2019	To examine the automated discovery techniques regarding a realistic hybrid simulation of logistics using BPMN process model.	Discovery algorithms were found to perform better with more extensive event logs.
[11]	Knoll et al., 2019a	To propose a methodology that integrates Multidimensional Process Mining (MDPM) techniques with principles of lean production and VSM and apply and evaluate it in a German automotive manufacturer.	This study provided an MDPM methodology for steady recording, assessment and waste analysis of each process within internal logistics.
[12]	Intoyoad et al., 2020	To propose a sequence clustering methodology based on Markov chains and expectation-maximization and validate it using real-life event logs from three different companies in manufacturing and logistics.	The proposed methodology can improve process discovery when confronted with dynamic and complex business processes with a satisfying process model quality.
[4]	Intoyoad and Becker, 2018b	To propose a methodology to improve the limitations of process mining by using a Markov chain as a sequence clustering technique in the data preprocessing step to extract the business process models using an experiment with real-world event data.	The proposed methodology can improve the quality of discovered process model by the measurement of replay fitness dimension.
[13]	Piest et al., 2021	To propose and evaluate the use of the Open Trip Model (OTM) for process mining in logistics in an operational logistics scenario.	The preliminary results are promising and demonstrates how the basic requirements for process mining can be fulfilled, but the support requires further experimental research and comparative studies.
[14]	Terlouw, 2017	To analyze the logistics process at a higher level of abstraction using the DEMO methodology and applied it for several organizations.	Provide better understanding on why things go wrong and what to do about them.
[3]	Becker and Intayoad, 2017	To improve the automated mining by adding context information to the process detection approach using process frequency and cycle time-based method. The processes used are lists of consecutive operations in a manufacturing environment.	Best results are achieved by selecting the most frequent processes as main processes and using them as additional contextual information.
[15]	Knoll et al., 2019b	To propose an extended ontology focusing on process mining within internal logistics using the methodology of Noy and McGuinness (2001) and PROMPT methodology and applying them in one production plant of a high model-mix assembly line in automotive industry.	Developing an internal logistics ontology for preprocessing.
[16]	Prathama et al., 2019	To explore the potential of the trace clustering approach to detect concept drift using a case study in a logistics process of a company in Indonesia.	Partition-based trace clustering could be used to understand the concept drift.
[17]	Pane et al., 2021	To visualize the effect of process mining in the context of logistics using design science research approach in a large international manufacturing company.	Process mining application contribute to an unbiased and more accurate analysis of the performance and compliance of the flow of materials and individual locations within that flow.
[6]	Benthuis et al., 2019	To create a mapping plan using heuristics miner algorithm to solve manufacturing and logistics problems in a company in Indonesia.	Each of the business processes provides time efficient and accurate decisions, resulting in project implementation comparable to the company’s business strategy.
[8]	van Cruchten and Weigand, 2018	To present a holistic process mining architecture to analyze and evaluate emergent behavior resulting from agent-based decision making using numerical experiments.	Workflow of a process mining model can be used to enhance the agent-based system, specifically, in analyzing bottlenecks and improving decision-making.
[18]	Schuh et al., 2020	To present a process mining approach for describing data-based process performance to easily map end-to-end processes using a real industry case study.	Emphasized the high potential of the presented approach and promise significant contribution to increase process performance when focusing on end-to-end order processing.
[2]	Intayoada and Becker, 2018a	To discover the relationship between situational context information and process lead time using datasets from three different manufacturing companies.	New methodologies and innovative concepts are required to assist process mining to deal with the dynamic environment.
[19]	Bemthuis et al., 2021	To define and examine bottlenecks and bottleneck classification levels using a model for classifying bottleneck analysis techniques in a logistics case study.	Proposed classification model is validated with the application of process mining bottleneck analysis techniques to a logistics case study.

reveals that the largest number of authors, which is 13, are from The Netherlands and they shared in six of the selected papers. The 11 authors from Germany shared in seven papers, which is the highest number of selected papers. In addition, there were six authors from Indonesia who shared in only two of the selected papers. However, two authors from the Czech Republic shared in one paper, two authors from South Korea shared in one paper too, and only one author from Sri Lanka shared in one paper.

Moreover, Figure 3 demonstrates the percentage of selected research papers in each of the six countries (i.e., Germany, The Netherlands, Czech Republic, South Korea, Indonesia, and Sri Lanka) that process mining in logistics was discussed in their research institutions. Accordingly, the selected research studies gave evidence that the area of process mining in logistic process of manufacturing organizations was researched and applied through case studies. As illustrated in Figure 3, the highest percentage of research papers (39%) was conducted in Germany, followed by the The Netherlands (33%).

Furthermore, the selected studies were analyzed on the basis of the purpose and methodology used in each study and its outcomes and contributions. All of these studies discussed process mining and how it can be effective with logistics processes. As described in Table 4, only two studies used the sequence clustering methodology and different methodologies were used in the other 14 studies such as BPMN process model as in [1], MDPM techniques as in [11], Open Trip Model as in [13] among others. Most proposed methodologies were implemented and their effectiveness was justified through case studies such as in [11] where the MDPM methodology was found to be useful for steady recording, assessment, and waste analysis of each process within internal logistics. Moreover, the classification model proposed in [19] showed that the application of process mining bottleneck analysis techniques is validated in a logistics case study. All of the selected studies used real case studies except for [1,8].

The selected studies were analyzed and evaluated in terms of several criteria related to process mining. These criteria included the presence of algorithms, modeling languages, and the types and tools of process mining in selected papers as illustrated in

Table 5. Analysis of process mining elements.

Ref.	Author and Year	Algorithms	Modeling Languages	Type of Process Mining
[1]	Halaška, Šperka 2019	Structure Heuristics Miner (sHM6), Split Miner (SM), Inductive Miner (IM), Fodina (FO), and α$	Petri Net	Discovery
[11]	Knoll et al., 2019a	Inductive miner	-------------	Discovery, Performance, and Conformance Analysis
[12]	Intoyoad et al., 2020	Heuristic mining, sequence clustering using expectation-maximization	Petri Net	Discovery
[4]	Intoyoad and Becker 2018b	Heuristic mining, sequence clustering	Petri Net	Discovery
[13]	Piest et al., 2021	Open trip model	------------	Enhancement
[14]	Terlouw, 2017	Inductive mining	-------------	Discovery
[3]	Becker and Intoyoad 2017	k-medoids clustering	ML vectors	Discovery
[15]	Knoll et al., 2019b	Know-Ont, unified foundational ontology	Process Specification Language	Preprocessing
[16]	Prathama et al., 2019	Trace clustering, inductive miner	Petri Net	Discovery and conformance checking
[17]	Pane et al., 2021	${\mathrm{P}\mathrm{M}}^2$ methodology, SAP, location sequence check, transformation algorithms	Data abstraction	Preprocessing
[6]	Bemthuis et al., 2019	Heuristic miner	XML, Python	Discovery, conformance checking, and enhancement
[8]	van Cruchten and Weigand, 2018	Alpha (α) miner, Integer Linear Programming (ILP), and inductive miner	Petri Net	Discovery, conformance checking, and enhancement
[18]	Schuh et al., 2020	Inductive miner	Petri Net	Discovery
[2]	Intoyoad and Becker, 2018	Naıve Bayesian classifier	Confusion, Matrix, probabilistic logic	Discovery
[19]	Bemthuis et al., 2021	Design science research methodology, classification	Petri Net	Discovery, conformance checking, and enhancement

. There are several algorithms that are used in process mining such as Alpha mining, Heuristics mining, and clustering. The modeling language depicts the language used in process representation and some of the languages used are Petri Net, BPMN, and UML. However, there are three types of process mining: discovery, conformance checking, and enhancement of business processes. Although process discovery starts with an event log to discover a process model, process conformance analyzes and evaluates how effectively an event log conforms to a designated process model. The concept behind process enhancement is to improve or extend the a priori process model by employing information from event log [20].

The analysis of the selected studies, as shown in Table 5, demonstrated the use of several algorithms. All studies show the use of one or more algorithm such as Alpha (α) miner, Heuristics miner, inductive miner, sequence clustering, and k-medoids clustering. This means that the use of algorithms in the selected studies reached 100%. The purpose of selecting these algorithms was depicted in

Table 6. Purpose of chosen algorithms.

Ref.	Author and Year	Algorithms	Purpose
[1]	Halaška, Šperka 2019	Structure Heuristics Miner (sHM6), Split Miner (SM), Inductive Miner (IM), Fodina (FO), and α$	To solve the problem of noise and incompletion of event logs, identify combinations of split gateways that capture the concurrency, conflict and causal relations between neighbors, discover duplicate activities, guarantee soundness and re-discoverability of discovered process models.
[11]	Knoll et al., 2019a	Inductive miner	To filter noise.
[12]	Intoyoad et al., 2020	Heuristic mining, sequence clustering using expectation-maximization	To deal with noise and reveal the main behavior of business processes.
[4]	Intoyoad and Becker 2018b	Heuristic mining, sequence clustering	To capture the frequency of the sequences, detect short loops and skipped activities, and deal with noise and exceptions.
[13]	Piest et al., 2021	Open trip model	To help logistics companies in The Netherlands share real-time logistic data efficiently.
[14]	Terlouw, 2017	Inductive mining	To discover the most frequent activities and process paths and the dependencies and time between different activities/events.
[3]	Becker and Intoyoad 2017	K-medoids clustering	To cluster the processes.
[15]	Knoll et al., 2019b	Know-Ont, unified foundational ontology	To develop an internal logistics ontology for preprocessing of data within process mining.
[16]	Prathama et al., 2019	Trace clustering, inductive miner	To group process trace according to the similarities criteria on a sequence of activities, called profiles; to construct a process tree for given log.
[17]	Pane et al., 2021	${\mathrm{P}\mathrm{M}}^2$ methodology, SAP, location sequence check, transformation algorithms	To support their internal logistic and business processes, check the compliance of the workflow, transform the data into the desired location perspective format.
[6]	Bemthuis et al., 2019	Heuristic miner	To overcome noise and provide fast results.
[8]	van Cruchten and Weigand, 2018	Alpha (α) miner, integer linear programming (ILP), and inductive miner	To remove all traces that are not fully processed yet.
[18]	Schuh et al., 2020	Inductive miner	To describe process performance and for its robustness.
[2]	Intoyoad and Becker, 2018	Naive Bayesian classifier	To assess the connection between process behavior and relevant context information.
[19]	Bemthuis et al., 2021	Design science research methodology, classification	To check the maturity of bottleneck analysis techniques and how far the analysis has reached.

. The inductive miner was the most common algorithm used in the selected papers. Six studies used inductive miner, representing 40% of all papers. Few of these studies did not mention the modeling language used to represent the process. It can be seen that 12 out of the 15 studies mentioned a modeling language, representing 80%. Petri Net was used in 7 studies, representing about 47% of all studies. However, several other languages were used such as Python, XML, ML vectors, process specification language, and data abstraction.

Furthermore, Table 5 included the types of process mining used in the selected studies. Most of these studies addressed the discovery part of the process. Table 5 shows that seven out of the 15 studies used process discovery as the main issue to be addressed representing 47%. However, five more studies addressed discovery in addition to other types of process mining such as conformance and enhancement. Moreover, one study addressed enhancement alone and two studies dealt with the preprocessing phase of process mining. The tools were not mentioned in all of the studies. Only eight out of the 15 studies mentioned the tools used, representing 53.3%. Three of these studies used DISCO as a process mining tool and the other five studies used ProM, ProM 6, ProM 6.6, ProM 6.7, and ProM Lite 1.1.

4. Conclusions

This paper introduces a structured review of the available literature about process mining in logistics from the year 2004. In this systematic review, fifteen published studies were selected and analyzed. The findings of this review show that all of the selected studies used process mining models, with thirteen of them based on real case studies. The studies employed various process mining techniques and algorithms, with only three of them not mentioning the modeling language used. The most frequently used algorithm was the inductive miner, and the Petri net language was the most commonly used for process discovery.

According to the literature searched, there was no systematic review before this paper that addressed process mining implementation in logistics related to manufacturing companies. However, the manuscript acknowledges several limitations. First, the review depended solely on the availability of academic literature, potentially missing relevant non-academic materials. Second, the number of papers on process mining in logistics for manufacturing companies was limited, indicating a gap in research. Third, the review focused primarily on studies conducted in The Netherlands and Germany, which may limit the generalizability of the findings. These limitations should be addressed in future research.

The manuscript suggests several areas for future research. Firstly, the authors recommend exploring missing published papers to fill gaps in the literature, including nonacademic materials. Additionally, they propose developing new approaches for process mining applications in logistic processes, given the limited number of papers on this topic Furthermore, the manuscript suggests conducting more systematic review studies on process mining in logistics, expanding beyond manufacturing companies.