Ontology‐Driven Formal Conceptual Data Modeling for Biological Data Analysis

Abstract

Biological data modeling serves many purposes, and many approaches exist that are used in this endeavor. The main topics of advanced conceptual data modeling for database and Object-Oriented software development to support biological data analysis are included in Figure 1.1, which extend the traditional 'waterfall' software development methodology as depicted in bold in Figure 1.2. The scope of this chapter is to provide an overview of the ontological and logical aspects of conceptual data modeling tailored to molecular biology and biological knowledge discovery.Many databases and software applications have been and are being developed in bioinformatics, which, following good computing methodologies, are-or should have been-developed in stages, going from requirements analysis ('what should the envisioned software do?') and conceptual analysis ('what data should it be able to manage?') to design-level code and then to the actual implementation. It is well-known that omitting the conceptual analysis stage by going straight to coding or scripting just adds to the pile of one-off (bioinformatics) tools that have more bugs and are much less, or not at all, maintainable and interoperable. Conversely, availing of a proper software development methodology with a representitle, edition.

Conceptual Model Interoperability: A Metamodel-driven Approach

“…A different strand of research on unification that does not use a metamodel, is to use one logic formalism for several conceptual modelling languages, notably a Description Logic language [8,13,15]. Different logics are used, however, and they do not cover all features of the language due to the complexity trade-offs made.…”

“…Different logics are used, however, and they do not cover all features of the language due to the complexity trade-offs made. For instance, in [8] identifiers are absent, and the DLR ifd used in [15] does not consider the ORM's relationship constraints or UML's aggregation. Also, approximate transformations are not represented.…”

“…Besides one-off unidirectional algorithms to transform a language, e.g., from ORM to UML [5], several multi-language approaches exist, ranging from linking each model to a graph [7] or description logic language [15] to transformations mediated by a dictionary of common terms [3]. However, these solutions are only partial, for they, among others, omit several constructs (e.g., weak entity types, roles) or modify the language (e.g., by removing datatypes from UML), and therewith have imprecise 'equivalence' mappings or the algorithms are not available.…”

Toward an Ontology-Driven Unifying Metamodel for UML Class Diagrams, EER, and ORM2

“…Concerning unification by means of a single formalisation in a chosen logic, there are separate formalizations, which can be seen as prerequisites, and partial unifications, e.g., [1,5,9,19,22,23,27,32]. Their approach is, mainly, to choose a logic and show it fits sufficiently with one of the CDM languages, and perhaps due to this, different logics are used for different CDM languages, therewith still not providing the sought-after interoperability for either of the languages or among each other.…”

The Foundational Ontology Library ROMULUS

“…It has been 15 years since the introduction of the notion of ontology-driven information systems [1], which entails ontology-driven conceptual data modelling [2][3][4][5]. Ontology-driven conceptual data modelling uses principles and solutions from Ontology (philosophy) and ontologies (as artifacts) to improve the quality of a conceptual data model and refine its language, which therewith improves the quality of the information system.…”