Ontology-Based Model Abstraction

Abstract

In recent years, there has been a growth in the use of reference conceptual models to capture information about complex and critical domains. However, as the complexity of domain increases, so does the size and complexity of the models that represent them. Over the years, different techniques for complexity management in large conceptual models have been developed. In particular, several authors have proposed different techniques for model abstraction. In this paper, we leverage on the ontologically well-founded semantics of the modeling language OntoUML to propose a novel approach for model abstraction in conceptual models. We provide a precise definition for a set of Graph-Rewriting rules that can automatically produce much-reduced versions of OntoUML models that concentrate the models' information content around the ontologically essential types in that domain, i.e., the so-called Kinds. The approach has been implemented using a model-based editor and tested over a repository of OntoUML models.

Foundational ontologies, ontology‐driven conceptual modeling, and their multiple benefits to data mining

“…Another aspect related to which ontology‐driven modeling languages can contribute to domain understanding is through their mechanism to support complexity management (Figueiredo, Duchardt, Hedblom, & Guizzardi, 2018; Guizzardi, Figueiredo, Hedblom, & Poels, 2019). Complex domains require representations that are both large in scale, and rich in subtleties.…”

“…In this article, we take the position that foundational ontologies and the process of ontological analysis supported by them serve as a fundamental support for establishing grouping criteria and similarity calculation, reducing the possibility of creating groups of objects that do not reflect genuine real‐world regularities. Because they serve as a basis for identifying the essence of entities of a given kind (Guizzardi et al, 2019), foundational ontologies are potentially valuable for identifying similarities in clustering process that are not merely accidental. In this direction, the identification of the foundational categories from which the concepts are derived, makes it possible to determine their nature, thus elucidating the differences between, for example, objects and events, dependent and independent entities, kinds of things and their roles, among others.…”

Relational Contexts and Conceptual Model Clustering

“…So, instead of breaking down the model into clusters that correspond to what one could intuitively call sub-domains, that approach brakes down the model in terms of general ontological categories. In contrast, the approaches of [13] and [16,17] differ from the approach presented here since these are approaches for model summarization and hence lossy approaches. Finally, [16,17] also differs from our approach since it requires user input in selecting a set of entities in the model that are of particular relevance.…”

“…3). There are three CM-CM methods in the literature that are based on the same language, namely, the approaches of (i) [6], (ii) [13], and (iii) [16,17]. The first method is the one that is closer to work presented here, since it is also a clustering method and, hence, a lossless approach.…”

“…For models that are centered on taxonomic relations (e.g., product types, biological taxonomies), we recommend alternative complexity management techniques, in particular, the static ontological views as proposed in [6]. In fact, this paper can be seen as a companion to [6] and [13] in a general research program of defining ontology-driven complexity management theories, techniques and tools. While in these two papers the focus is on model recoding with ontology-design patterns, and on model abstraction, respectively, here we propose the notion of relationship-centric conceptual model modularization (or clustering).…”