Source-linked AI summary
Data-Related Challenges and Requirements for Event Log Generation in Process Mining: A Systematic Literature Review
Ghita El Alaoui Talibi, Oleksandr Kosenkov, Anastasija Nikiforova
TL;DR
Event-log generation lacks a systematic mapping between data-related challenges and requirements, limiting efforts to identify gaps, prioritize improvements, and design solutions. This systematic literature review identifies 29 challenges and five requirement groups that address practically all of them.
Problem
A systematic mapping between data-related challenges and requirements for event log generation is lacking, limiting gap identification, prioritization of improvement efforts, and solution design.
Source: §1 · p. 2Method
Results
The review identified 29 challenges and five requirement groups addressing practically all challenges, except the effort- and time-consuming nature of event log generation.
Source: §6 · p. 11Takeaways & Limitations
Available event-log-generation solutions already apply data-engineering and domain-knowledge-engineering techniques, but not systematically.
Source: §8 · p. 13The mapping between challenges and requirements relied on the authors’ interpretation because primary studies lacked consistent, explicit descriptions of their interconnection.
Source: §7 · p. 13
Abstract
from arXiv · showhide
Process mining is becoming an essential technology that not only supports the improvement of business processes but also enables the software technologies underlying the future digital economy. Among the multiple challenges discussed by the process mining community, data quality is emerging as the most pressing one. However, insufficient attention has been paid to event log generation as the very first phase of process mining and to its relationship with data engineering and requirements engineering. To address this gap, we conducted a systematic literature review of data-related challenges, requirements, and solutions in event log generation. Our results show that a wide range of 29 challenges can be addressed with 5 basic categories of requirements. Existing solutions already embed some form of data engineering and domain knowledge engineering techniques for event log generation. This review can support researchers and practitioners in understanding current trends in event log generation for process mining and in leveraging data and requirements engineering techniques to improve log generation and, ultimately, process mining outcomes.
1 Introduction
The paper reviews data-related challenges, requirements, and solutions in event log generation, addressing the gap in systematic mapping between challenges and the requirements needed to tackle them. It focuses on improving data quality early and connecting process mining with data engineering and requirements engineering.
1 Introduction
- The lack of a systematic mapping between event-log-generation challenges and their requirements limits researchers’ and practitioners’ ability to identify gaps, prioritize improvements, and design effective solutions.The process mining manifesto identifies discovering, integrating, and cleaning event data as major unresolved challenges, while existing studies have proposed solutions for log-generation data quality.Source: §1 · pp. 1–2
- The review systematically synthesizes challenges, requirements, and solutions to help identify important or under-addressed topics and requirements for event log generation.It aims to help researchers and practitioners navigate the literature and determine which requirements should be implemented to address particular groups of challenges.Source: §1 · p. 2
- The study prioritizes addressing data-quality challenges during event log generation rather than relying only on post-hoc log cleaning and refinement.It argues that logging and process mining are often studied separately, and that process-aware logging should be designed with process-mining requirements in mind.Source: §1 · p. 2
- The study examines how data engineering and requirements engineering can contribute methods for addressing data-related challenges and defining process-mining-suitable logging requirements.Requirements engineering can establish data requirements for process mining and logging requirements for software systems; the review also identifies research gaps and existing contributions from both disciplines.Source: §1 · p. 2
2 Related work
Prior studies address process-data quality and event log generation, but none provides a systematic mapping of data-related challenges to requirements. This gap leaves practitioners without an integrated basis for proactive, quality-focused logging.
2 Related work
- Study identifies process-data quality as a key frontier but does not map specific data-quality challenges to requirements for implementing quality-focused logging systems.It examines issues such as determining whether the right data are collected and preventing process-data quality problems.Source: §2 · pp. 2–3
- Study [26] overviews event-data extraction, correlation, and abstraction techniques, but lacks systematic study selection and focuses on solutions rather than the challenges they address.Source: §2 · p. 3
- Study [38] surveys practical conceptual and technical challenges in event log generation but covers only a subset and provides no systematic categorization.Source: §2 · p. 3
- The 12 logging guidelines in were not empirically derived or linked to specific challenges, and are at least partly outdated given process-mining advances.Source: §2 · p. 3
- Study categorizes healthcare event-log-generation challenges by process step and reports corresponding solutions, but offers no higher-level synthesis or cross-domain perspective.A systematic overview of challenges and requirements could provide a baseline for more systematic reporting in future empirical studies such as.Source: §2 · p. 3
- Existing studies do not systematically map data-related event-log challenges to requirements, leaving research fragmented and practitioners without guidance for prioritizing proactive logging improvements.The review also identifies the connection between process mining and data engineering and requirements engineering as understudied.Source: §2 · p. 3
3 Background
Event log generation transforms raw system data into business event logs suitable for process mining, distinguishing process-focused business logs from system-focused logs. The study frames this work through data engineering and requirements engineering.
3 Background
- Business event logs capture process execution, unlike system event logs focused on system behavior, and are generated from raw event data and system logs for process mining.Logging can also support correctness tracing, debugging, failure handling, system analysis, and recovery.Source: §3 · p. 3
- Requirements engineering systematically elicits, models, and analyzes stakeholder needs, constraints, and domain information to define and achieve software-system goals.[47]Source: §3 · p. 4
- In this study, data-related challenges concern data processing during event log generation, while data-related requirements are those implemented in that process.Source: §3 · p. 4
4 Methodology
The study uses a systematic literature review and meta-synthesis to identify data-related challenges and requirements in event log generation and examine how data and requirements engineering techniques address them. Its methods define three research questions and document the search, screening, extraction, and synthesis process.
4 Methodology
- The review followed Kitchenham’s guidelines and software engineering empirical standards [50], using meta-synthesis to address three questions on challenges, requirements, and engineering techniques.The search was scoped to recent process-mining research, with the study noting commercial adoption and market growth in 2018–2019 as reasons recent studies could better represent current practices and challenges.Source: §4 · p. 4
4.1 Search strategy
- The search strategy used Scopus and IEEE Xplore, iteratively refined and benchmark-validated a query, and retained the final query after trial snowballing found few additional papers.The query was (“event log” OR “event data”) AND (“generation” OR “extraction” OR “logging”) AND “process mining”; it included all benchmark studies.Source: §4.1 · pp. 4–5
4.2 Study selection and screening
- Study selection began with title-and-abstract screening and continued with full-text review.The criteria were applied during both stages.Source: §4.2 · p. 5
Inclusion Criteria
- Studies had to be in English and address process-mining-relevant event data logging, data quality during logging or transformation, event log generation, or related requirements and engineering techniques.Source: Inclusion Criteria · p. 5
Exclusion Criteria
- The review excluded non-peer-reviewed publications, work on event-log refinement or repair, model-based log generation or model extraction, and synthetic-log generation or unrelated logging.Source: Exclusion Criteria · p. 5
Search and Screening Process
- The screening process reduced 561 initial records to 59 papers selected for full-text data extraction, after duplicate removal, date cutoff, title-and-abstract screening, and full-text review.The initial search had no publication-time filter; papers published before 2019 were excluded during screening. Each selected paper was read in full, with details recorded manually using an Excel-based extraction form.Source: Search and Screening Process · p. 5
Data analysis and synthesis
- Researchers used two rounds of coding to harmonize labels, mapped challenges to requirements through author interpretation, and classified solutions by their relation to data or requirements engineering.Because few studies reported distinct data-quality details, those mentions were integrated into the requirements category.Source: Data analysis and synthesis · p. 6
5 Results
The review identifies 29 data-related challenges in event log generation and groups requirements into five categories. It also reports data engineering techniques and domain knowledge as relevant to addressing these challenges.
5.1 Data-related challenges in event log generation for process mining (RQ1)
- Among 29 challenges, data quality (39%), data integration (27%), and non-process-centric data (24%) were most frequently reported; quality issues include missing, imprecise, inconsistent, incorrect, and irrelevant data.Data integration is difficult because event-log data often resides across systems, while non-process-centric data may lack a process-oriented structure.Source: §5.1 · p. 6
5.2 Data-related requirements for event log generation for process mining (RQ2)
- The review reports data-element availability as a frequently identified process-mining requirement, listing timestamps (26 studies), case IDs (23), activities (19), events (10), and labels (10).Relationships between events, activities, and cases were also mentioned as important for understanding process flow.Source: §5.2 · pp. 7–8
- Requirements call for accurate, unique, complete, trustworthy, relevant, consistent data with suitable granularity, as well as accessible data and understood database structures.Structures may need schemas or mappings for objects, events, and attributes, plus primary- and foreign-key constraints.Source: §5.2 · p. 8
- Clear definitions, event categorization, verified mappings, and domain knowledge support accurate assignment of data to event-log components.Domain knowledge was mentioned as a requirement in 17% of studies and can be represented through vocabulary mappings, rules, models, ontologies, or abstraction tables.Source: §5.2 · p. 8
Techniques overview
- Event-log preparation techniques are grouped into extraction, correlation, and abstraction, with standards, metrics, and one LLM-based study among reported solutions.Extraction identifies event data across heterogeneous sources; correlation groups data by process instance; abstraction maps data to activity events and addresses granularity differences. Standards mentioned include XES, XOC, and OCEL; the LLM study generated labels and recommended connectors from natural-language prompts and queries.Source: Techniques overview · p. 10
Data engineering techniques in process mining
- Reported data engineering techniques include ETL, scripts, Power Query, SQL Server, and Python, while advanced examples include complex event processing, distributed processing, and data federation.Some studies call for tools with data-engineering capabilities that support process analysts who are not data experts.Source: Data engineering techniques in process… · p. 11
Requirements engineering
- The reviewed studies do not directly report requirements engineering techniques, but domain knowledge matters across extraction, correlation, and abstraction.Extraction draws on domain concepts and relationships; correlation depends in part on choosing a suitable case notion; abstraction methods may use process models or other domain information. Some approaches involve domain experts to document knowledge and help maintain logs as domains or systems change.Source: Requirements engineering · p. 11
6 Discussion
The discussion maps 29 event-log-generation challenges to five requirement groups, while identifying effort and time as the sole challenge without a specific requirement. It highlights complementary roles for data engineering and requirements engineering in addressing data, process, and domain-knowledge needs.
6 Discussion
- Five requirement groups address 29 identified challenges, with effort- and time-consuming generation (Ch7) the sole exception.The authors suggest that most event-log-generation challenges are known and requirements for addressing them are established.Source: §6 · p. 11
- Classical data challenges, including data quality and missing elements, map to developed R1 and R2 requirements, but guidance for contextualizing them by domain or project is lacking.Source: §6 · p. 12
- Data-processing challenges such as integration and locating data map to R3, sometimes alongside R4, and data federation could help resolve integration challenges.Source: §6 · p. 12
- Domain-knowledge challenges mainly call for R4, sometimes R5; process scoping can require R3, R4, and R5 together.Source: §6 · p. 12
- Some challenges stem from software-system logging, including missing data, granularity, variability, standardization, and formatting, and can be addressed through R1 and R2 guidance.The authors propose recommendations to developers on data elements and relevant quality attributes.Source: §6 · p. 12
- Existing solutions reuse data-engineering techniques for extraction and address domain-knowledge challenges where requirements-engineering techniques may apply.Source: §6 · p. 12
- Requirements engineering could support logging best practices and log quality, capture initial process-mining requirements and scope, and model domain knowledge with verification, validation, and quality assurance.Logging implementation depends largely on developer expertise; one primary study proposed analyzing source code to identify relevant fragments in legacy systems.Source: §6 · p. 12
- Only one selected primary study focused on LLMs, which the authors suggest could help address event-log-generation challenges, including domain-knowledge availability.Source: §6 · p. 12
7 Threats to validity
The review used repeated author checks and two analysis rounds to address validity threats, but mapping challenges to requirements remained interpretive because primary studies reported their connection inconsistently. Independent iterative review, discussion, and an open mapping dataset were used to address this subjectivity and support transparency.
7 Threats to validity
- Both authors reviewed search trials, checked search results against gold-standard papers, discussed extraction and intermediary analyses, and used trial snowballing to assess its need.Analysis proceeded in two rounds: the first author categorized the data, then the second author reviewed the analysis and suggested changes.Source: §7 · pp. 12–13
- Because primary studies lacked consistent, explicit descriptions of challenge–requirement connections, their mapping relied on author interpretation and was independently reviewed in two iterative rounds.The authors discussed and resolved discrepancies and ambiguities.Source: §7 · p. 13
- An open dataset documents the challenge–requirement mapping reported in the studies, and the authors encourage more consistent, explicit reporting in future work.The dataset is intended to enhance transparency and reproducibility.Source: §7 · p. 13
8 Conclusion
The review identifies five basic requirement groups for addressing event log generation challenges and finds that engineering techniques could contribute more systematically to this foundational process mining phase.
8 Conclusion
- Event log generation transforms raw event data and system logs into business event logs, and five basic requirement groups can address its wide range of reported challenges.The review treats this as an initial and fundamental phase of process mining.Source: §8 · p. 13
- Available solutions already apply data engineering and domain knowledge engineering techniques, but not systematically; requirements engineering could guide logging functionality suited to event log generation and process mining.Data engineering could help resolve extraction and integration challenges, while requirements engineering could address domain and knowledge engineering challenges in scoping, mapping, and analyzing event logs.Source: §8 · p. 13
- The review recommends closer exploration of intersections among process mining, requirements engineering, software engineering, and data engineering, whose roles span domain knowledge, logging quality, and data processing and integration.Requirements engineering concerns domain and system-operation knowledge; software engineering implements logging functionality and supports log quality; data engineering processes and integrates data for process mining.Source: §8 · p. 13