Open Notebook Ontology development

One of the challenging aspects of ontology development is that in the early stages there are likely to be major collisions of orientation, and chemistry is no exception. So we need to prepare for some hard work addressing these. Traditionally this has been done by getting together physically, identifying the issues and working through them on the hope that a useful amount of agreement could be reached. It’s still extremely useful to meet and discuss.

However the web provides a different process where we can be much more inclusive and share our world views at an early stage. The plus side is that everything is visible; the downside is that without some process to orient thought it can become chaotic. To that end Nico Adams has been logging his current ontologies and thoughts:

ChemAxiom: An Ontology for Chemistry – 1. The Motivation

ChemAxiom: An Ontology for Chemistry 2. The Set-Up

ChemAxiom: An Ontology for Chemistry 3 – Choosing an Upper Ontology

I am not even going to start to summarise as Nico has been and will continue to do this. But I’ll add some comments. In posting his ideas to the web Nico is effectively doing Open Ontology development. I am delighted that Michel Dumontier – who also has a chemical ontology – has picked this up – within hours – and has offered to collaborate – both on Nico’s blog and on Twitter. Because we realise that ontologies are hard, we know that by combining ideas and infrastructure we are more likely to create something that reaches a wide community, is useful and can be explained.

We all have a major challenge in chemical ontologies. Most chemists have never heard the term and – reaslistically – most will only be mildly interested in the idea (many will see it as irrelevant). So we need to work out cases where this can be seen to be uaseful and more than an abstract philosophical exercise. There is less problem with bioscience – there the concept of ontology is commoner and also seen to be necessary in bringing together different disciplines.

By carrying out this design in public Nico gives us the chance to get widespread input, but this carries the certainty of colliding world views. What are these likely to be? I see at least:

  • Whether or not to use an upper ontology and if so which. Here we are guided by bioscience which has already – in many cases – chosen to do so and to use the Basic Formal Ontology, and Nico has done likewise – I trust his judgment.
  • Differences between bioscience and physical science. This will be one of the first full ontologies developed from a physical science point of view. There will certainly be differences in approach – chemists tend to have a more Platonic view of the world and their pragmatism leads to a fairly concrete and often object-oriented approach – nouns rather than verbs. Moreover much physical science is heavily based on consistent mathematical models and there is an underlying idea of a consistent set of axioms. By contrast bioscience is more phenomenological and categorical (I hope that’s right) where we try to impose some structure on a world that may have little other than our own.
  • Words. Words help to bring us together but also separate us if their basis varies between communities. What is a “measurement”? Is it a process, the result of a process, or a mathematical “point”? If we know how something is carried out in the lab that may affect our understanding of a word – is “sequence” a verb or noun? In general I find nouns much easier than verbs and CML concentrates on nouns (resulting in an object-oriented program system) and leaves relationships to other techniques such as RDF and ontologies.

So we have to look for consensus but also recgnise where there are fundamental irreconcilable differences. It will be hard, but it should be rewarding.

This entry was posted in Uncategorized. Bookmark the permalink.

2 Responses to Open Notebook Ontology development

  1. Frank says:

    I would encourage you to look at OBI ( during your ontology development as it is one of the largest ontologies using BFO and you will probably find there is some overlap in terminology and maybe some design patterns that will help. Also, are you aware that ChEBI is also going to be re-positioned under BFO?

  2. Nico Adams says:

    Hi Frank,
    thanks for the hints and encouragement. We have spent a while looking at OBI and there are clear areas of overlap and useful design patterns there – we completely agree….
    And yes, we are aware of ChEBI’s impeding refactoring too…thanks again for the helpful hints!

Leave a Reply

Your email address will not be published. Required fields are marked *