Michel Dumontier – Ontogenesis

Review for What is an upper level ontology?

Michel Dumontier — Wed, 12 May 2010 18:36:15 +0000

This is a review of What is an upper level ontology?

In this article, Robert nicely covers the different aspects that upper level ontologies need to consider to prescribe a coherent view of the world for its adopters.

Here are some specific comments that need to be addressed:

1. Instead of using electron as an example, i would use something more concrete – like a car.

2. Explain the notation for ” )1,2)”

3. Elaborate on abstract entities – why are these important, and give some examples.

Review of What is an ontology?

Michel Dumontier — Fri, 22 Jan 2010 13:31:49 +0000

This is a review of What is an ontology?

This well written article spans both logical and philosophical considerations in Ontology so as to provides insight into the kinds of entities that are believed to exist and how we might formally represent them and the basic relations that may exist between them.

The discussion on relations relating to identity (transformedInto, derivedFrom) necessitates further explanation. Are the criterion for identity embedded in physical continuity or in the conscious self? Indeed, we observe that a from develops from a tadpole, the idea lies in the material *largely* persisting spatiotemporally, and that the gain and loss of parts (and the corresponding qualities) is gradual and acceptably identity-preserving. Yet, we wonder whether the addition of even a single atom to a molecule through some chemical reaction maintains identity. To what extent does the gain or loss of parts become sufficiently important that it requires the distinction of forming a new entity? Perhaps more challenging is if we were to replace a person’s brain with another, we might perceive them to be the same individuals throughout the operations, but would this criterion for identity change if consciousness followed the brain? Then what might we say of identity? Important questions indeed for formal ontology and the representation of biological knowledge.

Review of Semantic Integration in the Life Sciences

Michel Dumontier — Fri, 22 Jan 2010 11:36:24 +0000

This is a review of Semantic Integration in the Life Sciences

This article discusses semantic data integration to address the overwhelming challenge of integrating and querying across thousands of biological databases, in a way that should be superior with “syntactic” data integration. It discusses at the highest level the kinds of data integration strategies: i) local-as-view where queries are mapped to their local sources (no transformation required) and ii) global-as-view where source data is transformed into a common schema. While ontology presents a salient opportunity to unify various data sources through a shared conceptualization, at least one example (protein, as it means for biopax/uniprot) demonstrates that this will be a significant, and non-trivial challenge. This is a great motivating example – it exemplifies syntactic (URI differences) and semantic (natural language/axiomatic definitions) heterogeneity -> how do we resolve this is a question worth of careful analysis in this article.

From schemas to ontology

The article needs to be reformulated in such a way that it clearly presents the problem of data integration and clearly defines and contrasts syntactic and semantic approaches, but more importantly, identifies the role of ontology in this process, as opposed to schemas. Ontology, in this sense, refers not only to an enhanced logic-based formalism in which class descriptions can be logically evaluated for equivalence through subsumption/consistency checking, but also to philosophical ontology such that different kinds of entities, including relations, can be integrated across domains.

From RDF to OWL

While the vast majority of RDF-based data integration efforts are fairly trivial (and require much time and effort), these do not exploit the explicit semantics found in rich ontologies. For instance, Bio2RDF ((Belleau F et al. J biomed inform 2008) now contains over 5 billion linked data statements using RDF, but there with no overaching ontology, queries must be formulated by tracing a path against existing resources. In contrast, data integration projects such as the pharmacogenomics of depression project (Michel Dumontier and Natalia Villanueva-Rosales. Briefings in Bioinformatics. 2009. 10(2):153-163.) use expressive logic-based ontologies that build on foundational philosophical ontology.

Vision

The vision, then, is that data integration occurs across domains through logic + philosophical ontology.

Ontogenesis: Who’s here?

Duncan Hull — Fri, 22 Jan 2010 09:04:10 +0000

Who’s here? The following is an alphabetical list of people currently attending the Ontogenesis Blogging a Book Experiment.

Sean Bechhofer, University of Manchester
Michel Dumontier, University of Carleton
Mikel Egana-Aranguren
Frank Gibson
Matthew Horridge, University of Manchester
Duncan Hull, EBI
Simon Jupp, University of Manchester
Allyson Lister, Newcastle University
Phillip Lord, Newcastle University
James Malone, EBI
David Osumi-Sutherland, University of Cambridge
Helen Parkinson, EBI
Robert Stevens, University of Manchester
Christopher Brewster, Aston Business School
Alan Rector, University of Manchester
Ulrike Sattler, University of Manchester
David Shotton, University of Oxford