Trait Ontology Workshop
Summary of Trait Ontology Workshop held in Montepelier, France, November 5-7 2009
TraitNet along with CNRS, France, organized a meeting to discuss developing a
trait ontology. Nineteen scientists met in Montpellier, France, November 5-7,
to discuss, design and develop a ‘Functional Trait Ontology’ for ecologists. As
the majority of ecologists present studied plants, it was decided that the
ontology would initially focus on plant traits. The ontology would be developed
by merging plant trait concepts listed in two categories, plant entities and
plant characteristics. Plant entities are the physical parts of the plant, and
plant characteristics are either direct, or derived, measurements made by
ecologists/biologists of plant entities. A semantic web-based application tool
developed by the group would be used to coordinate the clarification of the
definitions, and define a hierarchal structure and relatedness between the
plant trait concepts. It was decided to build the ontology using the OWL
language rather than the OBO standard. Besides the ontology, it was decided
that a semantic based registry of databases relevant to ecologists would be
developed to showcase the usefulness of working with data in a semantic
framework. The group has also decided to publish 2-3 papers related to the
scientific output that emerges in 2010.
Five major themes were to be discussed during the workshop:
- What should be the scope of the ontology? e.g. Do we include community and/or ecosystem level?). Also, relatedness to other fields (taxonomy, vegetation relevés, soil and climate environments…)?
- How does a trait-based ontology relate to existing ontologies in ecology?
- Identify the relevant concepts in the domain knowledge of traits.
- Discuss the best available ontological framework.
- Technical choices to implement the ontology.
With respect to ‘the scope of the Trait Ontology’ (Theme 1), the group concurred that to begin with “the Trait Ontology should be limited to plant functional traits, while being able to accommodate both expansion to a broader ecological knowledge domain and integration with other related ontologies”. Also found to be important was the need to identify and formalize ontological interfaces with other relevant ecology disciplines like taxonomy, climate, soil etc.
It was decided that a priority first step in developing the ontology was to discover what semantic knowledge existed of the plant traits in existing ontologies—such as the Crop Ontology (CO), Plant Ontology (PATO) and Trait Ontology (TO)—so as to optimize the reuse of semantic knowledge and structure (Theme 2).
It was decided to use the OBOE (Madin et al. 2007, Ecological Informatics, 2, 279-296) as a general framework to develop the plant functional trait-based ontology. With respect to identifying relevant concepts related to plant traits (Theme 3), it was decided that it would be best to split traits into groups: 1) physical plant entities, and 2) plant functional traits, i.e. traits that can be derived from plant trait entities. It was also decided that choosing the most appropriate ontological framework/model to define the plant trait knowledge domain would have to wait till after a few different inference models were tested in the following months (Theme 4).
With respect to choosing the best technological solutions for this project (Theme 5), it was decided that Isabelle and Marie-Angélique would lead this effort. An application being developed by Marie-Angélique would be used as the main tool to aggregate information regarding choosing relevant different trait entities and characteristics, and their respective definitions, and OWL was chosen over OBO as the language that the ontology would be built in.
Back (L-R): Shahid Naeem, Evan Weiher, Farshid Ahrestani, Jens Kattge, Bernard Amiaud, Sophie Gachet, Brian Enquist
Front (L-R): Eric Garnier, Daniel Bunker, Mark Schildhauer, Isabelle Mougenot