BFO Meets Critics

Panel Session at the Medical Informatics Europe Conference, Madrid, May 27-29, 2015

This panel will take the form of an author meets critics session prompted by the publication of the book Building Ontologies with Basic Formal Ontology by MIT Press in 2015. The panel will feature Barry Smith, a principal developer of BFO and one of the authors of this book, together with three prominent figures in the biomedical ontology field. Smith will present a brief summary of the book, which will be followed by critical remarks by the other panelists. It will conclude with Smith’s responses to these criticisms followed by questions to all panelists from the audience.

• Speakers:

Robert HOEHNDORF(Computer Science, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia)

Victor MAOJO (Artificial Intelligence, Polytechnic University of Madrid, Spain)

Stefan SCHULZ (Institute of Medical Informatics, Statistics and Documentation, Medical University of Graz, Austria)

• Respondent: Barry SMITH (Philosophy and Biomedical Informatics, University at Buffalo)
• Moderator: Ronald CORNET (Medical Informatics, Academic Medical Center, Amsterdam, Netherlands)

1. Introduction

Since 2006 the Basic Formal Ontology (BFO) has been serving as upper level architecture in a large number of biological and biomedical ontology initiatives as part of a strategy to enhance interoperability of data and information systems. Starting with the Open Biomedical Ontologies (OBO) Foundry, and now extending to include the Neuroscience Information Framework (NIF) Standard Ontologies, the IDO suite of Infectious Disease Ontologies, the cROP Common Reference Ontologies for Plants, and the VIVO-ISF (Integrated Semantic Framework) Ontologies, BFO is supporting the creation of what is intended to serve as a robust landscape of interoperable ontologies in biomedicine.

2. Ontological realism

BFO is built on the basis of a realist methodology, and many elements of this methodology are incorporated into the ontology itself, including the distinction between the types or universals which serve as the objects of scientific theory, and the instances of these types which are observed in the lab or clinic. This realist methodology has been used in turn as the basis of a series of criticisms of the so-called ‘concept orientation’ in medical terminology research, as also of a series of influential initiatives in the field of medical standards and ontologies, including HL7 RIM, SNOMED CT, and the National Cancer Institute Thesaurus. As documented by Schulz, et al., these criticisms have contributed to a paradigm shift away from what are called ‘concept representations’ and towards ontologies in the medical terminology field. At the same time, some have criticized the realist methodology, claiming for example that it leads to overcomplexity, or that it is superfluous. The proposed panel is devoted to such criticisms. It takes as its starting point the publication of a new book-length introduction to BFO and beginners’ guide to ontology building in biology and biomedicine, which will be published by MIT Press in early 2015. The publisher’s summary of this work is provided below.

3. Panel organization

The panel will begin with a brief summary of the book by Smith, to be followed by criticisms from the panelists, concluding with responses from Smith and with contributions from the audience. It is designed to be of general interest to biomedical informaticians, including those who have little technical knowledge of ontology. We will work to ensure also that the panel has a certain theatrical quality. The panelists will be invited to formulate their criticisms in a way that is maximally provocative. Smith will not receive notice of their criticisms in advance.

4. Short biographies of panelists

• Ronald Cornet publishes on a variety of topics in medical informatics, ontology, description logic, and knowledge representation, and his works play an important role especially in the study of medical terminology systems.
• Robert Hoehndorf is a prominent contributor to ontologies in biology, and has published papers also in knowledge representation, semantic integration, semantic interoperability and phenotype informatics. His Leipzig thesis of 2009 introduces the biological core ontology GFO-Bio, which grew in turn out of the General Formal Ontology which is in some respects an early version of BFO.
• Victor Maojo has worked since 1996 on biomedical database integration, most prominently in the field of clinical genomics. He has also contributed seminal papers in the new field of nanoinformatics. Other research areas include artificial intelligence in medicine, clinical guidelines and protocols, image processing and analysis, data and text mining, information retrieval, internet-based applications, models and tools for clinical trials and semantic interoperability. In 2011 he and his colleagues published an influential criticism of BFO and its associated methodology.
• Stefan Schulz has published a series of critical papers on BFO and related topics, including the role of realism in medical terminology research. He is also a leading figure in SNOMED CT research, in medical text and data mining, and formal methods in ontology.
• Barry Smith has published many papers in ontology.