Business Process Management (BPM) is a discipline encompassing a wide array of methods, techniques, and tools derived from both Information Technology and Management Sciences to manage business processes within organizations. This paper addresses a critical BPM challenge, specifically concerning the semantic enrichment of process data. We propose a novel framework designed to augment Object-Centric Event Logs (OCEL), a recent advancement in Process Mining, with the robust ontological foundation provided by the Basic Formal Ontology (BFO). By integrating OCEL data with BFO, our framework seeks to enrich process mining artifacts with deeper semantic meaning, thereby enabling more sophisticated analysis, more comprehensive understanding of business processes, also opening the possibility to obtain richer insights from Process Mining algorithms.
The Business Process Management (BPM) discipline provides a structured approach to understanding,
optimizing, and controlling the business processes that operate in enterprises [
The field of Process Mining, which systematically utilizes event data to improve business processes [
To tackle this gap between knowledge and data, this paper addresses the challenge of augmenting the
event data stored in OCEL logs with process and domain knowledge. To achieve this, we propose a novel
framework that integrates OCEL data with ontologies, specifically leveraging the Basic Formal Ontology
(BFO) [
This paper is structured as follows. Section 2.1 introduces the BPM discipline and the Basic Foundational Ontology (BFO). Section 3 presents the OntoOCEL framework that augments data from OCEL logs with BFO concepts. Section 5 concludes the paper, outlining directions for future work.
Business Process Management (BPM) is the discipline that includes methods, techniques and tools from
Information Technology and Management Sciences to support the design, analysis, configuration,
execution and monitoring of business processes [
The notion of business process is central to BPM. A business process consists of a set of activities that
are executed to handle a case (a process instance) [
Taking advantage of this event data and process models, Process Mining is the discipline that
systematically uses event data to improve business processes [
The fact that only procedural knowledge is tracked in BPMS system has been already identified
as one of the major unresolved issues in BPM [
Recently, Semantic Web technologies started gaining increasing attention for their ability to promote heterogeneous data integration and interoperability. Ontologies provide a common vocabulary for representing knowledge and information across heterogeneous resources and applications.
To support such interoperable environment, diferent types of ontologies exist [
There exists a number of upper-level ontologies already in place, such as DOLCE, UFO, GFO, BFO,
etc [
BFO top-level ontology has been designed with a realist instance in mind. Its main goal is to use
ontologies to represent the knowledge acquired as a result of scientific eforts. Being realist means that
such knowledge captures general features of reality as general theories (i.e. generalizations and laws
of science), rather than particular facts. To make such distinction, BFO presents some fundamental
categories as described below:
Universal, Particular and Defined Classes. Universals are mind-independent entities that can
be repeatedly instantiated across time and space, with an indefinite number of particulars, while
particulars are individual entities, restricted to specific places. They instantiate universals, but cannot
be instantiated. Defined classes are general terms used in science to refer to particular individuals in
reality (e.g., medical doctor, dog) that have no corresponding universals [
BFO is structured in terms of two disjoint hierarchies of universals, depending on how particulars relate to time. Continuants and occurrents are the roots of each branch: Continuant and Occurrent: Continuants endure through time, retaining their identity, fully existing at any time they exist. Examples include a house, an apple, color of an orange. In contrast, occurrents unfold over time, being composed of temporal parts. Examples include a talk, a race, the history of Brazil, a period of time in which the sun rises, people attending a meeting.
Process and Temporal instant: A process is an occurrent composed by some temporal proper parts, a time interval = [, ..., ] in which a material entity is as participant, while a temporal instant is a zero-dimensional temporal region with no temporal parts.
In scientific research, it is relevant to categorize diferent types of information entities that carry scientific knowledge. For this reason, the Information Artifact Ontology (IAO) is an ontology created using BFO as a basis, capturing numerous information entities existent in scientific research (e.g., protocols, documents, experimental logs, databases, published literature and so on). Information Content Entity. In IAO, an information content entity is a generically dependent continuant that refers to (is about) some entity.
Document. A number of information entities that must be understood together as a whole.
As an upper ontology, BFO provides the foundation for over 350 domain ontology extensions in
multiple domains [
This section describes the OntoOCEL framework to augment OCEL event data with process and domain knowledge:
The first step of our methodology consists of conducting a semantic analysis to understand how the process knowledge structure can be mapped to the to upper level ontology, regardless process instances.
In BPM literature, a business process is composed of a set of activities executed to handle a case
(Section 2.1) (the control-flow perspective). Activities and processes are performed by roles that can
be fulfilled by people or organization units (resource perspective), manipulating data and information
(data perspective), happening during a period of time (time perspective) [
An assessment of BFO classes (Section 2.2), together with use cases on how to relate particulars to
BFO universals and defined classes defined in [
Now that we know how the process knowledge relates to the upper ontology, we can think about the domain knowledge. To do this, we initially look into the data stored in the event logs.
To facilitate interoperability among process mining tools, event logs are standardized. Figure 2(a)
depicts the OCEL metamodel adapted from [
With the OCEL event logs in hands, one can understand the data objects that we have available data, together with the events that manipulate such data. These data objects will be our domain entities in the next step.
Example. With the event logs in hands, it is possible to identify two data objects specified within the log ( and ) and one activity ( ) □
The third step consists of understanding the nature of knowledge stored in these data objects and how they fit with the knowledge structure of process knowledge (and indirectly to BFO). Activities are instantiated as subclass of BFO process, the data objects are instantiated as subclasses of BFO information content entity, roles are instantiated either as subclass of (defined class) person or as subclass of BFO object aggregate. The data objects also carry the values of the data object attributes. These are modeled as OWL Properties.
Example. Figure 3(a) depicts the hierarchy of BFO classes in OWL. First, the OCEL object types ( , and ) are instantiated as subclasses of information content entity. As one can see in the hierarchy, BFO provides diferent types of subclasses as carrier content (e.g. ifgures, documents, email, etc.). Here, the class has been created as subclass of BFO document and pointed to class (subclass of BFO object), since a BFO information content entity is about some entity in reality. Relations among concepts are modeled as OWL Object Properties in Figure 3(b). Activities , and are created as subclass of BFO process. □
In this step, the knowledge structure of the enterprise may be complemented with declarative knowledge.
This knowledge should be related with the procedural knowledge modeled in the previous steps.
Example. Imagine an admission process that patients may have diferent types of diseases. The
doctor would like to know about it, he wants to track it along time. In this case, we insert a
and as a disposition as depicted in [
This step concerns the linkage between processual and domain data with the ontology. The Cellfie
plugin [
The idea of providing semantics for BPM technologies has been embraced in two veins in literature.
A first group of approaches use Ontologies and Semantic Web technologies combined with business
process models [
In a second strain of research that augment event data with Semantic Web technologies [
This paper has presented a comprehensive framework, OntoOCEL, designed to address the critical semantic gap in Process Mining by augmenting Object-Centric Event Logs (OCEL) with a rich ontological foundation derived from the Basic Formal Ontology (BFO). We have outlined a five-step methodology that facilitates this integration, moving from a semantic analysis of process knowledge in relation to upper-level ontologies to the development of a domain-specific ontology, its complementation with declarative knowledge, and finally, the seamless integration of event data with the constructed ontology.
While the current work lays a conceptual and methodological foundation, it is important to acknowledge that populating the ontology with event data is a limitation of our work that must addressed in future eforts. Future work will focus on the empirical validation of this framework through the actual import and analysis of real-world OCEL event logs, by elaborating the transformation rules, and exploring the potential for automated reasoning and intelligent decision support based on the enriched ontological models. This will ultimately lead to more intelligent, adaptive, and insightful BPMS systems.
During the preparation of this work, the author used ChatGPT-4 in order to: improve writing style. After using this tool, the author reviewed and edited the content as needed and take full responsibility for the publication’s content.