Some take-away's from my recent encounter with the Phenotype Research Coordination Network

Biosphere 2


As I'm departing from Biosphere 2 from an extremely enjoyable and productive meeting of The Phenotype Research Coordination Network, I'm reflecting on my lessons learned from diving into this research community I was rather unaware of previously. 

Foremost on my mind is that I have found bioinformatics researchers for which high quality ontology modeling is a standard means of addressing knowledge organization, sharing and integration - to the extent that the benefits of using ontologies do not have to be questioned any longer. At the same time, also unquestioned, is the benefit of deep i.e. expressive formal semantics and also of formal deductive reasoning over this semantics. I cannot recall having found this combination previously in such a strong way: high-quality ontology modeling (for sharing and reuse) coupled with using OWL reasoners at (or beyond) the limits of their capabilities.

The other prominent observation which lingers on my mind is how similar this community's main obstacles wrt. the use of semantic technologies are to those in other application communities I've interacted with - and how much some of these main obstacles are in fact not discussed prominently at major meetings of the core semantic web research field. I'm mentioning some of them in the following - and of course these are topics of concern for semantic web researchers, it just seems to me that they are not given the emphasis they would deserve in terms of what the big obstacles are for application areas.

  • How to make it easier to reuse and integrate ontologies (and their underlying data): How to deal with an ecosystem of overlapping heterogeneous ontologies and different needs of granularity and coverage, seamlessly.
  • Ontology alignment beyond the straightforward.
  • More and better tools for building high-quality ontologies and ontology based applications.
  • Making ontologies more accessible: Modularization, simplified views (without giving up the complex models), seamless integration of controled vocabularies.
  • More education and educational material. Concretely the question was: why do Computer Science departments in the U.S. generate so few graduates with Semantic Web expertise?
  • Development and working out of showcases for added value.

Finally - the absence of one topic which is currently (still) hyped among semantic web researchers was rather noticable, for me: Linked Data. In fact, I think that I actually was the only person at this meeting who even mentioned linked data, and that was only during coffee breaks. Of course (large!) OWL ABoxes were everywhere, as was RDF and SPARQL and Wikidata etc. But not - Linked Data as such.

My thanks go to the people running the Phenotype RCN for inviting me, and to Anne Thessen and Katja Schulz, who brokered the contact.