A number of conceptual modeling approaches have been developed from system and conceptual perspectives, adding logic to systems modeling and using ontologies to span the gap between systems and business semantics. The aim of this post is to bring under a common roof the main concepts proposed by these approaches and the ones identified by the Caminao Kernel (CaKe).
Behind lexical discrepancies, the underlying semantics of most of proposed approaches are congruent for a number of key concepts, in particular when defined in relation with UML (Unified Modeling Language). For simplicity sake two approaches are considered as representative, respectively of the Object-oriented and Goal-driven perspectives.
Object-oriented Perspective: UFO
Individuals & Universals
The Unified Foundation Ontology (UFO) is based on the ontological distinction between individuals and universals:
- Individuals are objects that can be uniquely identified in reality.
- Universals are sets of features which can be realized jointly by sets of individuals.
That classical philosophical understanding relies on the implicit assumption that the reality is physical, a questionable premise for the modeling of today’s social environments and business activities. That makes room for a conceptual blind spot when the symbolic/physical distinction is crossed with the one between extensional and intensional representations:
- Extensions represent individuals independently of their nature.
- Intensions represent universals independently of their realization.
Along that reasoning physical intensions would be associated with digital objects (as illustrated by digital twins), and symbolic extensions would be accounted for in terms of relationships between individuals. Applied to enterprises’ contexts and concerns, that understanding raises practical as well as theoretic issues:
- From a practical point of view, the need to cover the blind spot induces unwarranted complexity and circular references; for example, contracts (symbolic), would have to be defined in terms of relationships between organizations (symbolic), themselves defined in terms of contracts between persons (actual).
- From a theoretic point of view, modeling languages should always remain neutral and never coerce semantics into preformatted meanings.
Hence the need to reconsider the way things are identified.
Endurant & Perdurant Types
Terms may vary but things are commonly categorized in terms of endurant and perdurant types.
Endurant types describe things that can be identified independently of time, perdurant ones deal with phenomena, typically events and processes.
Substantial types describe endurants with their own identity principle, moment types are meant to denote the manifestations of substantial endurants, relational or intrinsic. They coincide with CaKe’s aspects: the former with connecting nodes, the latter with features.
The blind spot mentioned above hinders a clear and balanced representation of substantial endurants, with a core of endurants (sortal types) with clearly defined semantics, and leftovers (nonsortal types) only defined through predicates.
As it happens, the way sortal and nonsortal substantial types are defined suggests a formal and agreed upon distinction between extensional and intensional categories.
Extensional (or structural) categories represent sets of endurants with homogeneous identity principle whatever its basis: biological, social, or mechanical. Intensional (or functional) categories represent sets of endurants with heterogenous identity principle.
CaKe’s specialization semantics are defined accordingly: subset of individuals for extensional categories, extend of aspects for intensional ones.
It must be noted that perdurant types can also be categorized in terms of extensions (actual occurrences of events and processes) and intensions (aspects of events and activities).
Goal-driven Perspective: i*/Tropos
Besides commonly agreed concepts (cf. i* and the Requirements Conundrum), the i*/Tropos methodology is characterized by its understanding of actors and goals.
Actors, Roles, Agents
Tropos defines actor as an abstraction (or concept) set into three dimensions: organizational (realized through positions), behavioural (realized through roles), and intensional (realized through agents).
Introducing an abstraction level between terms commonly used with overlapping semantics puts Tropos at cross-purposes with most other modeling approaches. Instead, the Caminao kernel tries to align typical semantics with contexts and concerns set at the same level of abstraction:
- Agent, active (with behaviours) entity with intentions (goals), identified in the environment
- Roles, sets of behaviours as defined by an organization
- Actor, subset of role as defined with regard to systems (UML)
Besides the ecumenical benefits, these definitions draw a line between system and enterprise architectures, putting the focus on the significance of agents and goals at the nexus between systems, enterprise, and business conceptual models.
Intentions & Goals
Intentions and goals, generally ignored by systems modeling languages, become critical for enterprise and business models. But, as should be expected, agreeing about EA semantics is a challenging endeavour; to mention a few issues:
- How to set apart intentions and goals defined at an individual level, from the ones defined collectively by organizations ?
- Could software agents have intention, or even goals ?
- How to trace and represent subgoals across decision-making processes ?
- How to weave together goals, knowledge, and accountability ?
Given the ubiquity of software agents and machine learning capabilities, these issues are critical ones, and so are the ways intentions and goals are represented. Yet these questions are mostly ignored, the focus being put on the representation of goals and an abstract understanding of agents.
With regard to goals, the main advance of Tropos is its distinction between hard- and soft goals: the former explicit and therefore open to planning, monitoring, and assessment; the latter for supporting rationales.
That distinction can be aligned with CaKe’s business and use cases:
- Business cases define objectives at the enterprise level independently of specific business processes
- Use cases define objectives at the enterprise level with regard to specific business processes
- System cases describe the direct contribution of systems to business processes
As for agents and roles, the objective of the Caminao approach is to maintain a clear distinction between systems and enterprise representations, using rationality and accountability to draw the lines:
- Goals (or objectives) are attached to agents; but instead of intentions, they are defined in terms of rationales and decision-making.
- Rationales can support the behaviours of all kinds of agents: machines, people, or organizations; but decision-making can only be carried out by accountable agents, namely people or organizations
- Individual decision-making can rely on implicit or explicit knowledge; but collective decision-making can only be supported by explicit knowledge.
Customary connectors like dependencies and means-ends analysis can be directly mapped using CaKe connectors.
The primary objective of conceptual modeling languages is to span the whole of enterprise and business contexts and concerns, typically:
- Understanding the flows of data as observed in environments
- Modeling the information managed by systems, the associated business processes, and the enterprise organization
- Managing knowledge of business environments and objectives
Taking into account the diversity of domains and agents, and the changing nature of competitive environments, a seamless integration of these dimensions can only be achieved through open-ended and ecumenical solutions.
- EA in bOwls (Overview)
- Modeling Paradigm
- i* and the Requirements Conundrum
- People & Personas
- Knowledge Architecture
- Views, Models, & Architectures
- Ontologies & Models
- On Pies and Skies: Abstraction in Models
- Conceptual Models & Abstraction Scales
- Models & Meta-models
- Ontologies & Enterprise Architecture
External READING (sample)
- Albuquerque, A. and Guizzardi, G. (2013), An ontological foundation for conceptual modeling datatypes based on semantic reference spaces. In IEEE 7th International Conference on Research Challenges in Information Science (RCIS), 1–12. IEEE.
- Amund Tveit, A survey of Agent-Oriented Software Engineering
- Angelo Susi, Anna Perini, John Mylopoulos. The Tropos Metamodel and its Use, Informatica 25
- KAOS modelling: a short introduction
- John Mylopoulos, Information Models and Conceptual Models
- Guizzardi, G. Towards an Ontological Account of Agent-Oriented Goals