Ebook: Formal Ontology in Information Systems

Since its inception ten years ago, the International Conference on Formal Ontology in Information Systems (FOIS) has explored the multiple perspectives on the notion of ontology that have arisen from such diverse research communities as philosophy, logic, computer science, cognitive science, linguistics, and various scientific domains.

As ontologies have been applied in new and exciting domains such as the World Wide Web, bioinformatics, and geographical information systems, it has become evident that there is a need for ontologies that have been developed with solid theoretical foundations based on philosophical, linguistic, and logical analysis. Similarly, there is also a need for theoretical research that is driven by the issues that have been raised by recent work in the more applied domains. FOIS is intended to be a forum in which to explore this interplay between the theoretical insights of formal ontology and their application to information systems and emerging semantic technologies. The papers appearing in this year's conference exemplify this interaction in very interesting ways, with papers covering the range from foundational issues and generic ontologies to methodologies for ontological engineering, reasoning, and ontology integration.

Themes emerging from the papers give us a snapshot of current issues within the fields of formal ontology and ontological engineering, as well providing us with a glimpse of future research directions.

Although ontologies were originally motivated by the need for sharable and reusable knowledge bases, the reuse and sharing of ontologies themselves is still limited because the ontology users (and other designers) do not always share the same assumptions as the original designers. It is difficult for users to identify implicit assumptions and to understand the key distinctions within the ontology and whether or not disagreements reflect fundamentally different ontological commitments. The challenge therefore still stands to propose ontological engineering methodologies that emphasize ontology reuse and that identify the characteristics of an ontology that enhance its reusability.

The interaction between ontology and science has emerged as an interesting new theme. First, ontologies can be treated as scientific theories, rather than as engineering artefacts. Second, there are ontologies for scientific theories, such as biology and chemistry, which play a role in integrating multiple data sets. Finally, there is work on developing ontologies of scientific theories; such ontologies play a supporting role in scientific research by providing explicit representations for alternative theories. This is also interesting insofar as one can consider ontologies for scientific domains to be a proposed solution for the sixth of twenty-three challenge problems posed by David Hilbert in an address to the International Congress on Mathematicians in 1900:

Mathematical treatment of the axioms of physics:

The investigations on the foundations of geometry suggest the problem: To treat in the same manner, by means of axioms, those physical sciences in which mathematics plays an important part.

The identification and logical formalization of fundamental ontological distinctions continues to be an impetus for current research. Fundamental distinctions, such as universals vs. particulars, features vs. substrates, and artefacts vs. roles are still a source of many challenging problems. Many of these issues converge in unexpected ways, particularly in the treatment of collective objects. Linguistic expressions have also motivated several areas in formal ontology, particularly in the areas of vague predicates and geographic terminology.

Ontology evaluation, both from a logical and empirical perspective, has also been recognized as a critical phase in ontological engineering. On the one hand, this leads to a deeper understanding of the relationships between ontologies. As ontologies are increasingly being deployed on the web, users are faced with the dilemma of selecting amongst multiple possible ontologies for similar domains. On the other hand, ontology evaluation is based on the relationship between the ontology and its initial conceptualization and intended application. We can rigorously characterize the relationship between the intended models of the ontology and the models of the axiomatization of the ontology, but it is more difficult to evaluate the correspondence between the intended models and their adequacy for the intended application of the ontology.

Finally, the widespread deployment of ontologies also raises the challenge of managing discrepancies that arise between ontologies. Although this is most evident in applications that require the integration of multiple domain (and possibly upper) ontologies, the problem must also be addressed by end users who want to merge concepts from multiple ontologies to create new ontologies that meet the specific needs of some domain. Integrating sets of independently designed ontologies also has ramifications for supporting automated reasoning with the ontologies.

The success of FOIS-08 has been the result of a truly collaborative effort. We would like to thank the members of the Programme Committee for their diligent work and constructive feedback, which have contributed to an excellent conference programme. We would like to thank the three invited speakers, Johanna Seibt, York Sure, and Mike Uschold, for providing their interesting perspectives on formal ontology in information systems. Finally, we would like to thank the Conference Chair, Nicola Guarino, and the Local Chairs, for managing all of the details that have made the conference a productive interaction of researchers from the diverse disciplines that contribute to formal ontology.

Carola Eschenbach

Michael Grüninger

We trace the roots of ontology-drive information systems (ODIS) back to early work in artificial intelligence and software engineering. We examine the lofty goals of the Knowledge-Based Software Assistant project from the 80s, and pose some questions. Why didn't it work? What do we have today instead? What is on the horizon? We examine two critical ideas in software engineering: raising the level of abstraction, and the use of formal methods. We examine several other key technologies and show how they paved the way for today's ODIS. We identify two companies with surprising capabilities that are on the bleeding edge of today's ODIS, and are pointing the way to a bright future. In that future, application development will be opened up to the masses, who will require no computer science background. People will create models in visual environments and the models will be the applications, self-documenting and executing as they are being built. Neither humans nor computers will be writing application code. Most functionality will be created by reusing and combining pre-coded functionality. All application software will be ontology-driven.

Increasingly, in data-intensive areas of the life sciences, experimental results are being described in algorithmically useful ways with the help of ontologies. Such ontologies are authored and maintained by scientists to support the retrieval, integration and analysis of their data. The proposition to be defended here is that ontologies of this type – the Gene Ontology (GO) being the most conspicuous example – are a part of science. Initial evidence for the truth of this proposition (which some will find self-evident) is the increasing recognition of the importance of empirically-based methods of evaluation to the ontology development work being undertaken in support of scientific research. Ontologies created by scientists must, of course, be associated with implementations satisfying the requirements of software engineering. But the ontologies are not themselves engineering artifacts, and to conceive them as such brings grievous consequences. Rather, ontologies such as the GO are in different respects comparable to scientific theories, to scientific databases, and to scientific journal publications. Such a view implies a new conception of what is involved in the authoring, maintenance and application of ontologies in scientific contexts, and therewith also a new approach to the evaluation of ontologies and to the training of ontologists.

The Ontological Square is a four-categorial scheme that is obtained by crossing two formal distinctions which underpin conceptual modelling languages and top-level ontologies alike: that between types (or universals) and tokens (or particulars) on the one hand, and that between characters (or features) and their bearers (or substrates) on the other hand. Thus the Ontological Square consists of particular substrates, called substances, and universal substrates, called kinds, as well as particular characters, called modes or moments, and universal characters, called attributes. In this article, I try to elucidate the basic ontological assumptions underlying this four-category scheme and I propose a calculus of many-sorted second-order logic that is meant to capture these intuitions by an enrichment of standard atomic logical form. A first-order semantics can be designed for such a Logic of the Ontological Square, with respect to which the latter's soundness can be established.

Formal ontology relies on representation languages for expressing ontologies. This involves the formal semantics of these languages which is typically based on a limited set of abstract mathematical notions. In this paper, we discuss the interplay between formal semantics and the intended role of ontologies as semantic foundation. In this connection a circularity is identified if ontologies are to determine the conceptual equivalence of expressions. This is particularly relevant for ontologies which are to be provided in multiple formalisms. In order to overcome this situation, ontological semantics is generally defined as a novel kind of semantics which is purely and directly based on ontological entities. We sketch a specific application of this semantics to the syntax of first order logic. In order to beneficially rely on theoretical results and reasoning systems, an approximation of the proposed semantics in terms of the conventional approach is established. This results in a formalization method for first order logic and a translation-based variant of ontological semantics. Both variants involve an ontology for their application. In the context of developing a top-level ontology, we outline an ontology which serves as a meta-ontology in applying ontological semantics to the formalization of ontologies. Finally, resolved and remaining issues as well as related approaches are briefly discussed.

The deployment of learning resources on the web by different experts has resulted in the accessibility of multiple viewpoints about the same topics. In this work we assume that learning resources are underpinned by ontologies. Different formalizations of domains may result from different contexts, different use of terminology, incomplete knowledge or conflicting knowledge. We define the notion of cognitive learning context which describes the cognitive context of an agent who refers to multiple and possibly inconsistent ontologies to determine the truth of a proposition. In particular we describe the cognitive states of ambiguity and inconsistency resulting from incomplete and conflicting ontologies respectively. Conflicts between ontologies can be identified through the derivation of conflicting arguments about a particular point of view. Arguments can be used to detect inconsistencies between ontologies. They can also be used in a dialogue between a human learner and a software tutor in order to enable the learner to justify her views and detect inconsistencies between her beliefs and the tutor's own. Two types of arguments are discussed, namely: arguments inferred directly from taxonomic relations between concepts, and arguments about the necessary and jointly sufficient features that define concepts.

Medical research and clinical practice deal with complex and heterogeneous data. This requires a systematic approach for semantic integration of information to support clinicians in their daily tasks.

As the clinicians speak and think in a very different language than that of the computer scientists, existing knowledge engineering approaches based on classical expert interviews fall short. Moreover, as human health is a very sensitive subject, the reuse of standardized hence reliable ontologies as medical knowledge resources becomes a key requirement.

In this paper, we first discuss the specific medical knowledge engineering requirements, we identified along a semantic medical image and text retrieval use case. Then we report on ongoing work towards establishing a corresponding methodology based on ontology reuse that is derived from the requirements. The methodology, which will be discussed in detail, relies on a novel technique for semi-automatically generating a set of potential user queries to support the knowledge elicitation process.

In the last years, several methodologies for ontology engineering have been proposed. Most of these methodologies guide the engineer from a first paper draft to an implemented – mostly description logics-based – ontology. A quality assessment of how accurately the resulting ontology fits the initial conceptualization and intended application has not been proposed so far. In this paper, we investigate the role of semantic similarity as a quality indicator. Based on similarity rankings, our approach allows for a qualitative estimation whether the domain experts' initial conceptualization is reflected by the developed ontology and whether it fits the users' application area. Our approach does not propose yet another ontology engineering methodology but can be integrated into existing ones. A plug-in to the Protégé ontology editor implementing our approach is introduced and applied to a scenario from hydrology. The benefits and restrictions of similarity as a quality indicator are pointed out.

We are surrounded by collective phenomena, with examples existing on many levels of granularity. Despite our frequent experiences of such phenomena they seem to have been largely ignored within the field of ontology. In this paper, existing ontologies are examined to determine the extent to which they can adequately represent collective phenomena, and are found wanting in a number of important respects. Our goal is to find good ways of representing collective phenomena in a way which does justice to the often subtle relationship that exists between the view of a collective as a whole and its constitution as a plurality of individual participants. An important prerequisite for this is to determine the range of variation that exists within the broad class of collectives. A number of example collective phenomena have been studied to extract appropriate classification criteria. The results from this study are used to produce a new typology of collectives which is intended to establish a basis for the adequate treatment of collectives within an ontology. The paper concludes with a set of further research questions that have been raised during the development of the classification.

The purpose of this paper is to examine different modelling strategies available in a multiplicative formal ontology, and the principles that drive their choice. This study is based on the results of recent work aiming at extending the foundational ontology DOLCE to grasp two quite different notions, that of artefact and that of role. These results, summarized in the paper, show that two multiplicative modelling strategies, entity stacking and property reification, are essential in both cases.

Resolving conflicts based on ambiguities in the public vocabulary is one of the challenges in semantic integration. Though different suggestions for resolving (ambiguity) conflicts with semantic integration operators exist, there is still a need for clear formalizations of adequacy criteria for the operators. In this article, adequacy criteria for semantic integration similar to rationality postulates of classical belief revision but adjusted to the semantic integration scenario are formalized. The criteria are intended to capture integration settings in which the integration candidates are well developed ontologies with a shared public vocabulary. In such cases, both ontologies have to be preserved in some form in the integration result and have to be recoverable from the integration result. Additionally, the integration result has to be consistent and provide connections between the integrated ontologies. The criteria are applied by evaluating a small collection of integration operators that solve conflicts deriving from ambiguities in the public vocabulary.

State of the art formalisms for distributed ontology integration provide ways to express semantic relations between homogeneous components of different ontologies; namely, they allow to map concepts into concepts, individuals into individuals, and properties into properties. However, the extensive usage of multiple distributed ontologies requires the capability for expressing different forms of mappings, which extend the semantic relations among homogeneous components studied so far. In recent papers extensions of the Distributed Description Logic (DDL) have been proposed to represent mappings between heterogeneous elements; i.e. mappings connecting concepts and relations. In this paper we investigate the computational properties of reasoning with mappings between homogeneous as well as heterogeneous elements in distributed ontologies, and an effective decision procedure for reasoning with multiple ontologies bridged with both forms of mappings.

In the last years, the vision of the Semantic Web fostered the interest in reasoning over ever larger sets of assertional statements in ontologies. It is easily conjectured that, soon, real-world ontologies will not fit into main memory anymore. If this was the case, state-of-the-art description logic reasoning systems cannot deal with these ontologies any longer, since they rely on in-memory structures.

We propose a way to overcome this problem by reducing instance checking for an individual in an ontology to a (usually small) relevant subset of assertional axioms. This subset can then be processed by state-of-the-art description logic reasoning systems to perform sound and complete instance checks for the given individual. We think that this technique will support description logic systems to deal with the upcoming large amounts of assertional data.

This paper presents our experience in reusing mereology ontologies in a Pharmaceutical Product ontology, an ontology built by the EU NeOn project. It shows a set of mereology ontologies implemented in different machine interpretable languages and analyzes them according to the different types of mereology identified by Varzi. Then, it describes the specifications of mereology modeling necessities for Pharmaceutical Product. Finally, it presents the ontology which fits best with the specifications. One of the results of this work is a procedure to reuse general (also called common) ontologies.

Chemical entities are the foundation of biochemistry and biology, but until now there have been few coherent attempts to produce a top-level ontology for chemistry to connect ontological descriptions of reality at the molecular level, such as ChEBI, with upper-level ontologies such as BFO, or indeed familiar laboratory-scale concepts such as mixtures. We work out relationships between chemical types that are compatible with the OBO Relation Ontology, describe macroscopic chemical systems in terms of grains and collectives, and propose a top-level ontology for chemically-relevant continuants and discuss it in relation to BFO and BioTop.

An ontology of general science knowledge (SKIo) is developed to enhance machine representation and use of scientific theories in emerging e-Science Knowledge Infrastructures. SKIo specializes the DOLCE foundational ontology with science knowledge primitives, such as science theory, model, data, prediction, and induction. These are arranged to reflect the complex knowledge structures used in science, such as scientific ideas playing different roles within and between theories. SKIo is encoded with OWL-DL, uses the DOLCE Descriptions and Situations module, and provides defining conditions for its primitives to enable an extensible bridge between DOLCE and domain science ontologies. An application to environmental theories is demonstrated, and its utility to other natural sciences is promising.

Effect axioms constitute the cornerstone of formal theories of action in AI. They drive standard reasoning tasks, especially prediction. These tasks need not be coupled with actual acting; the reasoning agent is, thus, typically given an ex post acto narrative of what actions took place. An acting agent, however, has no access to such knowledge; it needs to face what we call the event categorization problem, and figure out what actions it did. Until this is achieved, effect axioms will be useless. A careful review of the literature on effect axioms reveals that their syntax, semantics, and ontological commitments are so deeply entrenched in the armchair reasoning about action paradigm, that they cannot be used in resolving the event categorization problem. By enriching the ontology of action theories, we propose a different approach for representing effects of actions that unifies the two views. The enriched ontology is independently motivated by linguistic concerns.

The goal of the REMINE project is to build a high performance prediction, detection and monitoring platform for managing Risks against Patient Safety (RAPS). Part of the work involves developing an ontology enabling computer-assisted RAPS decision support on the basis of the disease history of a patient as documented in a hospital information system. A requirement of the ontology is to contain a representation for what is commonly referred to by the term ‘adverse event’, one challenge being that distinct authoritative sources define this term in different and context-dependent ways. The presence of some common ground in all definitions is, however, obvious. Using the analytical principles underlying Basic Formal Ontology and Referent Tracking, both developed in the tradition of philosophical realism, we propose a formal representation of this common ground which combines a reference ontology consisting exclusively of representations of universals and an application ontology which consists representations of defined classes. We argue that what in most cases is referred to by means of the term ‘adverse event’ – when used generically – is a defined class rather than a universal. In favour of the conception of adverse events as forming a defined class are the arguments that (1) there is no definition for ‘adverse event’ that carves out a collection of particulars which constitutes the extension of a universal, and (2) the majority of definitions require adverse events to be (variably) the result of some observation, assessment or (absence of) expectation, thereby giving these entities a nominal or epistemological flavour.