Process mining consists of the application of specialized algorithms to the event logs generated by information systems, to identify trends, patterns, and details about how processes are executed. In this sense, it constitutes a tool that combines data science and process analysis to discover, validate and optimize workflows.

By integrating data mining techniques with process analysis, organizations can extract and analyze the information contained in the event logs of their systems, thus achieving a more accurate understanding of the performance of their processes. This helps to detect bottlenecks, inefficiencies and critical areas that require improvement. Thus, the idea of Process mining is to discover, monitor and improve the real processes (i.e., not the superfluous processes)

Process Mining, based on a data-driven approach, favors objective decision-making regarding the allocation of resources and the prioritization of improvement initiatives. In this way, managers have concrete evidence to drive the transformation and optimization of existing processes (IBM, 2024).

The IEEE[1] published a Process Mining Manifesto, in an effort to move forward with its adoption (IEEE, 2011). The Manifesto introduces principles, intentions, concepts, and terms, defining Process Mining as a discipline that is located between computational intelligence and data mining, on the one hand, and process modeling and analysis.

The relationship between the different terms is indicated in Table 1.

Table 1: Relating the different terms.Source: Process Mining Manifesto

The available literature points to three types of Process Mining: discovery; compliance; improvement.

Discovery: creates process models directly from the event data.

Conformity: compares the actual processes with the intended model to detect deviations.

Improvement: optimizes existing models using additional information, identifying bottlenecks and efficiency opportunities.

Process Mining brings an objective and quantitative approach to Business Process Modeling (BPM), traditionally managed manually. While classic BPM is usually based on qualitative inputs obtained in workshops and interviews to then document workflows in process maps, Process Mining uses real event data generated by information systems. This makes it possible to accurately describe how processes are executed in practice, overcoming the limitations of subjective or informal approaches. As mentioned above, it applies only to implemented processes.

Why Process Mining is Important for Tax Administrations

Process Mining in tax administrations is applied for improvements in the following areas:

1. Transparency and compliance

Process Mining reconstructs and visualizes real processes from the event logs of computer systems.

In tax administration, this can help to:

Identify deviations from standard operating procedures.

Detect bottlenecks in workflows (e.g., taxpayer registration, audit cycles, returns processing).

Provide a traceability for internal and external supervision.

2. Operational efficiency

By comparing the actual workflows with the expected ones, tax administrations can:

Reduce processing times for declarations, requests, and payments.

Automating low-value stages.

Improve the allocation of resources.

3. Fraud detection and risk management

The combination of Process Mining and analytics allows:

Detect unusual routes in the processes that may indicate fraud, collusion, or errors.

Provide early warnings in high-risk cases.

4. Better use of existing data

Tax administrations already generate extensive event records in central tax systems, electronic invoicing platforms, ERP modules and document management systems, as well as in IT environment management systems. Process Mining turns this data into actionable intelligence without the need to collect added information.

5. Support for digital transformation initiatives

Process Mining complements:

Robotic Process Automation (RPA): By previously identifying stable and repetitive processes that are worth automating.

Business Intelligence (BI): By adding a temporal and procedural view to performance metrics.

Process Management: Objects and Events

In short, the objects represent the “what” or “who” is managed in the process (e.g., a taxpayer, a declaration, a penalty). Events, on the other hand, indicate “what happened” and “when” with respect to that object (e.g., statement delivered, payment received, audit performed).

The combination of both makes it possible to reconstruct the actual process and compare it with the planned model, facilitating the detection of deviations, inefficiencies, or risks.

Objects

Objects represent tangible or abstract entities within a process. Processes are structured around well-defined objects, which represent the entities on which activities and events are performed. These typical “objects” serve as the backbone for process modeling, Process Mining, and performance analysis.

The following classification is proposed for the tax objects:

1. Objects linked to the taxpayer

Taxpayer / Person / Legal entity: main register of the taxable person; Tax current account: consolidation of obligations, payments, balances, and tax credits; Unique tax identifier (NIF, RUC, TIN).

2. Objects of obligations and declarations

Tax returns (VAT, Income, Excise taxes, etc.); Electronic invoices and digital receipts; withholdings and receipts (retention agents, collection agents); Associated accounting records.

3. Objects of collection and payments

Payment orders / Payment receipts; Means of payment (transfer, card, bank window, digital wallet); Tax debt: principal, interest, penalties; Payment facility plans.

4. Objects of control and control

Control / audit cases; Notifications and requirements; Electronic files; Adjustment or sanction resolutions.

5. Objects of litigation and legal management

Administrative resources (claims, appeals); Judicial processes related to taxes; Administrative and judicial resolutions.

6. Objects of service and assistance to the taxpayer

Return requests; Requests for registration, updating or deregistration; Consultations / care tickets (call center, portals, chatbots); Certificates (tax compliance, tax residency, exemptions).

7. Internal management objects

Procurement and contracting processes; Human resources (payroll, tax career); Internal IT systems (incidents, developments, changes).

In Process Mining, these objects become instances or cases (cases) and allow them to identify the traces that make up the flow. For example:

Case/ Instance  “VAT Return #2025-08-87456”

Associated events  “Reception → Automatic validation → Crossing of inconsistencies → Notification → Payment → Closing.”

Events

An event is a well-identified occurrence in time and space, which belongs to a specific instance of a process and is recorded in an information system (Van der Aalst, 2016).

They allow us to compare the actual execution versus the planned model, detect deviations, bottlenecks, and opportunities for improvement.

The events have the following characteristics: connection to a case: each event is associated with a process instance (e.g., a taxpayer’s return or a refund request); executed activity: describes which process step or task occurred (e.g., invoice validation); time stamp: indicates the exact time when it occurred; resources or actors: optionally, identifies who or which system executed the action; additional attributes: such as monetary values, error codes, input channels, etc.

The data sources for capturing and describing the events, in the tax administrations, are typically the tax management information systems (SIGT or core tax systems); Electronic declaration platforms; Electronic invoicing modules and digital vouchers; Taxpayer assistance systems (CRM, call centers, chatbots); monitoring and control systems; Accounting and financial records; Documentary and electronic records management systems.

Database management systems (DBMS) and other IT environment management software packages also generate momentous events for certain business processes, which can be chosen to compose the sets of events required by Process Mining projects.

The necessary events are usually scattered over various Logs and can be consolidated into a single Log for analysis by Process Mining tools.

When consolidating these events, it is necessary to plan the following adjustments to ensure final quality: Unify date and time formats (including time zone); Eliminate duplicates and events with incomplete information; Correct impossible sequences (e.g., closing activity before starting activity); Standardize activity names (e.g., “Receipt statement” and “Receipt statement”); Align the case identifier between heterogeneous sources.

Platforms

Although interested tax authorities can gradually develop their own tools, there are powerful platforms on the market that speed up projects in this area.

Table 2 presents the conclusion of the evaluations conducted by the Gartner group, with its traditional “Magic Quadrant” applied to Process Mining platforms. More details in (Gartner, 2025).

Table 2: Gartner Magic Quadrant 2024 – Process Mining Rigs

Source: Gartner Group

Final considerations.

Although Process Mining is still a recent discipline, practical applications have already been detected in various fields, such as tax administration and software development.

In the City Council of Granada, citizen complaints about delays in the processing of files led to a review of internal procedures. To do this, Process Mining techniques were applied in the tax collection department, identifying bottlenecks and improvement points that were not obvious to employees or managers until then (Rozinat, 2020).

In the field of software development, established organizations depend on CI/CD pipelines[2] as the only way to bring new functionalities to production. Although the workflow seems simple in theory, in practice exceptions and deviations predominate. In this context, Process Mining emerges as a promising technique to detect breaches of standardized DevOps processes, identify bottlenecks and point out optimization areas in the software delivery chain (Nogueira & Zenha-Rela, 2024).

The future of Process Mining is projected towards closer integration with advanced technologies, enhancing its role in process optimization and digital transformation.

Bibliography

Gartner. (2025). Magic Quadrant for Process Mining Platforms 2024. Gartner Group. Available at: https://www.gartner.com/doc/reprints?id=1-2KRX9ZQJ&ct=250415&st=sb

IBM. (2024). What is process mining? Available at: https://tinyurl.com/bde5ur98

IEEE (2011). Process Mining Manifesto. Available at: https://www.tf-pm.org/upload/1580738243172.pdf

Nogueira, A., Zenha-Rela, M. (2024). Process mining software engineering practices: A case study for deployment pipelines. Information & Software Technology. Vol 168, Elsevier. Available at: https://www.sciencedirect.com/science/article/pii/S0950584923002471?via%3Dihub

Rozinat, A. (2020). Analyzing the Complaints Process at Granada City Council. IEEE Task Force on Process Mining. Available at: https://www.tf-pm.org/resources/casestudy/analyzing-the-complains-prociess-at-granada-city-council

Van der Aalst, WMP. (2016). Process Mining: Data Science in Action. Springer Berlin, Heidelberg. Details of the book: https://link.springer.com/book/10.1007/978-3-662-49851-4

References:

[1] Institute of Electrical and Electronic Engineers.

[2] Continuous Integration / Continuous Deployment

27 total views, 27 views today