Universal conceptual modeling: principles, benefits, and an agenda for conceptual modeling research

Most research reviewed converged on the notion that any universal form of communication, sign system, set of ideas, needs to be accessible, or easily understood, appreciated, and used. Empirically, the relationship between the accessibility features of ease of learning, ease of application, and widespread use has been established in education, linguistics, anthropology, psychology, as well as information systems studies.

Research shows that universal conceptual systems can be described using highly accessible and intuitive language and using accessible concepts. Anthropology, philosophy, and psychology, for example, studied universal concepts under the notion of basic-level categories. The research on basic-level categories suggests that there is a small list of “universal” concepts (e.g., “bird,” “tree”) [136], insofar as they are present in virtually every culture [11]. Furthermore, they are invariant to language proficiency and domain expertise: Both experts and novices understand the meaning of these categories and can use them effectively. They tend to be the first words patterns used when talking with children, and hence, the first basic words used by children [86]. Among other things, what makes the basic-level universal is its “accessibility”—its ease of understanding and use. The terms that express these concepts tend to be short, morphologically simple (e.g., car vs. sport utility vehicle). Furthermore, these terms map to what psychologists call “naturally occurring clusters” or “natural seams of the world” [112]: object referenced by these terms look alike and unlike the other “basic” objects in the same environment (e.g., trees look very different from birds, but somewhat similar to each other). This makes it easy to apply these concepts without creating much communication confusion. Notably, owing to these properties, basic-level categories have already been considered in conceptual modeling research [29], with the aim to promote usability of models, especially for novices.

Linguistics, anthropology, and psychology, in particular, investigate “linguistic universals” and “semantic universals,” the latter being defined as the properties of meaning shared by the languages of the world [58, 152]. These are terms, such as every, some, most, the, a, three, and present in virtually all languages and all cultures. Steinert-Threlkeld and Szymanik [152] provide an explanation of the presence of such universals due to their ease of learning, showing that expressions satisfying universals are simpler than those that do not. Notably, information technology scholars have developed a model of technology acceptance predicated on the ease of use (with learning being a component) of the technology [42, 164].

Related to the notion of accessibility is the theory of “universal design” [118], which states that design needs to be maximally inclusive and sensitive to all potential users. This means designs should address the requirement of marginalized communities on the grounds that if the design works for them, it should also work for others [118]. There are indeed many groups that are not considered as typical designers and users of conceptual models, such as IT novices [75], people with cognitive limitations due to age or illness, and visually impaired people (most models are visual in nature) [143].

Considering the research on accessibility and the aspirations of modeling for everyone, we propose:

Despite many perceptible differences between human languages, there are properties shared by all languages, including common core at the level of syntax, semantics, and phonology [7, 36, 76]. For example, despite a wide array of sounds across cultures, “universal phonetics” is present in all human societies, such as “all languages have stops” after the words and sentences [1, 76]. We already encountered one general explanation for this when dealing with basic-level categories: Commonality among humans is in part based on the mapping of the recurring patterns in the environment. Another mechanism behind these universals is the flexibility of form, as explained by the universal grammar theory.

A prominent group of linguists led by Noam Chomsky holds that underlying all natural languages have so-called universal grammar (UG)—innate principles that are instantiated when a speaker learns a particular language [35]. The UG theory describes a universal meta-language and defines the general principles and parameters (e.g., that every word can be identified with a linguistic category), which then become instantiated into specific languages. The UG theory suggests that a language aspiring to be general should permit large variation of expressions based on few principles and parameters. However, none of the principles and parameters are strictly binding, in that a given linguistic expression can deviate from them. Flexibility is desirable when it is impossible to predict exactly how a given language is going to be used by different people and in different situations. Among other things, flexible principles and parameters allow for great cultural diversity of human languages and permit for language evolution—all desirable characteristics for a language such as Datish.

Flexibility has also been suggested as a dimension of universality by other theories, most notably by theories of artifact design. For example, the theory of tailorable design suggests for tailorable technologies (flexible systems that can be modified by a user in the context of use) to be based on a dual design perspective (building some design functionalities, but allowing significant design choices to be defined by the user) [56]. Norman [118] promotes “universal design” and argues that “[t]he best solution to the problem of designing for everyone is flexibility: flexibility in the size of images and computer screens, in the sizes, heights and angle of tables and chairs. Allow people to adjust their own seats, tables and working devices.” In the same vein, Gregor and Jones [63] suggest a widely cited template for building design theories (theories of design and action). According to the authors, for design theories to be applicable to broader contexts, they need to be “malleable”—able to be modified, flexible.

Notably, flexibility is already considered in conceptual modeling research as a way to expand model uses and users. Hence, “flexibility” of conceptual models and the corresponding database schemas has been suggested as important to support highly heterogeneous user environments, such as social media, and crowdsourcing [97]. Similarly, flexibility in process modeling has been a design feature to support flexible organizational processes [39]. In the latter context, flexibility was defined as: “the ability of the workflow process to execute on the basis of a loosely, or partially specified model, where the full specification of the model is made at runtime, and may be unique to each instance” [141]. Flexibility through setting parameters and general intentions, rather than specific instructions, has been suggesting as a generally applicable approach to IT development [145]. This strategy is consistent with agile and DevOps methodologies [9, 87].

Underlying the flexibility prescriptions in design theories is the assumption that it is virtually impossible to predict all possible ways to interact with an artifact.³ Hence, by promoting flexibility and malleability of a conceptual model, more uses and users can be accommodated. In this sense, the theories of linguistics and design converge: to ensure systems of communication, or signs are widely used, they need to be malleable and flexible.

Flexibility stands in opposition to the common “closed world” assumption underlying many computing languages (programming, conceptual), where rules of the language are explicitly established and are strictly binding [133]. In contrast, we encourage flexible and creative use, adaptation, and evolution of UCM. As with UG and universal design, such openness and flexibility are fundamentally important for ensuring that UCM can cover a broad range of uses and is accessible to the widest audiences. This background motivates the following general principle of UCM:

In addition to accessibility and flexibility, universality is understood in reference theoretical foundations as ubiquity—widespread, high frequency of occurrence and use and commonality of the signs, concepts, and ideas.

As mentioned, psychology, anthropology, and philosophy study basic-level categories. One reason for the acquisition and use of basic-level categories is their ubiquity, or high frequency of occurrence in the general discourse. These arguments originate in the work of Zipf [173], who foreshowed the development of the basic-level theory by finding that more frequently used words tend to capture many of the terms needed for thinking and communication. Later, high usage frequency became a hallmark of basic-level categories [86]. Now, it is established that the relationship between frequency and universality is two-way: Frequent words, which tend to describe basic-level categories, are more readily available for acquisition and reference, which enhances the universality of basic-level categories [112].

Converging arguments and evidence from numerous domains, including linguistics, psychology, and economics, suggest that cognitive and linguistic forms that are more frequently occurring tend to be the most important socially and economically. These are captured by Zipfian or power-law distributions [116, 169]. Summarizing these effects, the physicist Buchanan [18] proclaims that ubiquity of something reflects a simple, but highly successful generative mechanism. Things in social or physical domains would not be ubiquitous unless they follow some fundamental universal laws [18]. Although we may not always be able to explicate these mechanisms, ubiquity can serve as a proxy for these universal laws or patterns.

The ubiquity principle is widely found in history and anthropology. For example, there is a small set of symbols nearly universally shared across cultures. Specifically, “shapes found across many cultures include lines, circles, spirals, zigzags, squares and squares of circles” [135]. These shapes have a relatively stable and common general and contextualized understanding (e.g., a circle in the night sky is commonly assumed to be the moon). Furthermore, the manner in which the symbols are drawn exhibit strong regularity within (e.g., by different age groups) and across cultures. For example, parallel sides and right angles are considered basic compositional forms for graphical elements, known as the “geometric regularity effect,” a recent discovery in science [44, 139, 140].

The ubiquity principle suggests the constructs, visual notation, and rules used in a universal language should be ubiquitous; that is, drawn from those frequently used by different people, in different situations and settings.

A corollary to the ubiquity principle is the existence of a core of elements, a small set of ubiquitous elements that account for much of communication and thinking. Another dimension of universality is minimalism—using the fewest possible constructs, signs, terms, visual symbols to communicate information.

The minimalism principle is a foundation of the generative systems theory. Generative systems (e.g., the number system, the Internet) have the capacity to produce both expected and unexpected variation as a result of open and fundamentally unpredictable use [157, 174]. According to the theory of generative systems, the single most important principle of a generative system design is minimalism or “procrastination” [31, 174]. The idea is to only endow a generative system with essential features for functioning and to permit these features to evolve or be extended by its users. This principle has a strong rationale: the presence of too many choices may create high cognitive load and impede the adoption and use of a universal system, especially by novice designers.

Minimalism has additional support in the limited cognitive resources of humans. Psychology research suggests humans are capable of operating with and memorizing seven, plus or minus two, chunks of information [109]. Hence, seeking to limit the essential number of constructs to between five and seven (or less) can facilitate better learning and use of a universal language.

Conceptual modeling languages seek to represent concepts about the world (cf. [127]). To be universal, these concepts (and therefore the models that contain them) should be the most basic, primitive ones, describing basic facts, and structures of the world.

We define primitivism as a principle of conceptual modeling that seeks to capture those concepts that are most fundamental, basic, and atomic, in terms of which other concepts are defined. For example, the concept of a business firm is not primitive as it requires the prior understanding of its constitute notions, including an organization, people, resources, association, legal authority, and liability. In contrast, the concept of resource is more primitive which is also underscored by the fact that commonly a resource is defined by providing examples of the resources, such as money, materials, and other assets (e.g., [45]).

Whereas minimalism dictates having a small number of elements, primitivism suggests for the elements to be foundational, or the atomic and indivisible elements upon which all other elements are based. The more primitive the concepts, the more applicable they are in generic contexts, rather than specialized discourse. Hence, the concept of business firm has a restricted domain of modern economy, whereas the more primitive concept of resource, has a much broader domain, encompassing not only economy, but also such domains as production and biology.

Many theories dealing with universal forms are based on primitivism and thus provide the foundations and guidance on how to achieve primitivism in conceptual modeling. The “linguistic universals” and “semantic universals” suggest there are foundational elements in communication and thought [58, 152]. Hence, basic linguistic terms (the, most) can be used to form more complex structures (e.g., the most). Geometric primitives, including point and line segment, are considered the foundational building blocks of visual forms [121].

Similarly, the cognitively universal basic-level categories are primitives in that they can be used to construct more complex cognitive structures [112]. For example, the basic (or primitive) concept of tree is the building block of a variety of more complex, or second-order concepts, such as deciduous tree, blooming tree, or native tree. Basic-level categories tend to be expressed with morphologically simple words, making it convenient to use them as a basic building block of verbal and written communication [136].

The idea of “ontological primitivism”—that there are basic building blocks of reality—can be found in philosophical theories about reality [10, 64, 68, 69, 102, 167]. Some philosophers believe the world can be reduced to its basic building blocks [21]. Others find an epistemological and methodologic justification: “every theory of nature needs to identify at least one ontological primitive, since we cannot keep on explaining one thing in terms of another forever” [82], p. 12.

Examples of ontological primitives include object, entity, or event. Different philosophers proposed lists of ontological primitives. Hence, Bertrand Russell [137] offered a complete “inventory of the world” consisting of “particulars, qualities, relations and facts.” For Ludwig Wittgenstein [171], a foundational element of reality was object. By far the most common philosophical approach is substance ontology [10, 48], which holds that the basic element of being is what is referred to as entity, object, thing, or substance. Some consider these terms synonymous. Halpin [70], for example, defines object as “any individual thing of interest.” For Bunge, object is the “undefined” primitive in terms of which other ontological constructs are defined. Conceptual modeling widely uses object as a key modeling construct. Object has been a core construct in conceptual modeling languages (e.g., UML, ORM, and OASIS) [70, 79, 91]. Object features as a key notion in a number of ontologies, such as UFO [65, 66], object-oriented ontology [71], and Basic Formal Ontology, BFO [149]. Hence, Guizzardi et al. [68] claimed “any reference theory for conceptual modeling would need a rich theory of entity (object) types.” Related concepts include entity, thing, system, these are present and are often considered primitive in a number of ontological theories (e.g., in BFO, UFO and Bunge), as well as foundational ideas of conceptual modeling [34, 70].

The concept is important in systems development more broadly (as in object-oriented analysis and object-oriented programming).

The principle of primitivism in theories of universals is accompanied by the principle of modularity. For basic building blocks to be combined to form more complex, composite structures, they need to have modular design. Modularity is the property of having distinct self-contained components that can be combined together to create new forms.

Design research on extensible, generative technologies converge on an important principle of universal, flexible, and accessible systems—modularity of system components [56, 174]. Modularity allows systems to be combined and recombined, based on evolving and individual needs. It is also present in many natural systems: “the universal design strategy [of life] involves ‘bottom-up’ synthesis and modular, hierarchical organization both within and across multiple length-scales” [84]. For example, hierarchical organization of human concepts permits advanced thinking and reasoning and is the basis of DNA [134]. The ontological postulate that the world is made of systems explains reality as organized in modular levels, whereby basic elements of reality are layered on top of one another (e.g., chemical system is the basis for biological system) [22].

Due to its centrality in nature, modular design is a common basis of engineering [84]. Modularity permits the management of complexity, which can still be a challenge, even when the number of design choices are intentionally minimized. Hence, the modularity principle can be enhanced by flexibility—reducing constraints on how the components can be combined.

Modularity is further important for UCM design because of the constraints the other dimensions place. Specifically, primitivism, minimalism, and ubiquity suggest limiting the number of constructs. This makes it difficult to ensure important information is represented in models. It also promotes general use at the expense of specialization. However, modularity enables recombining the foundational building blocks, thereby creating the possibility for more complex rules and ideas to be captured. Hence, modularity is an important element of natural languages, whereby basic linguistic elements can be combined and recombined, in practically infinite ways [35]. Other universal systems are modular, such as mathematics, where basic axioms can be used to derive more complex constructs needed for specialized uses. Thus, a logarithm can be derived from more basic operations (e.g., multiplication, power). The result is the ability of modular systems to support a wide range of more specialized thoughts and expressions.