The focus of my work is on definitions, and in particular, on predicting definition contents in specialized dictionaries and in ontologies. For a definition to be useful for computer scientists, people working with ontologies in data science, technical writers, scientists, legal experts, and translators (be it, for example, for annotating scientific texts or for translating specialized texts), it should convey the relevant information about the experts’ meaning of the defined term. For definitions to be useful for a computer system to reason upon the objects of a domain in an ontology, they might have to contain less information than what is relevant for human users.
In both cases, including the relevant kinds of information in the definitions is a common concern for non- or less-expert definition authors, such as terminologists and ontologists or domain experts. For example, a net boom in the domain of oil spill cleanup can be defined as ‘A boom to collect viscous oils at sea.’ or as ‘A boom that is made of netting.’, or both. The definition might even provide some other kind of information that might be relevant for using the term in the intended way, such as ‘by letting air and water pass through the net but not viscous oils’.
While ontologists and terminologists normally work in close collaboration with domain-experts who can tell them what is relevant, they are still often faced with the troublesome question of selecting the relevant kind of information for making definitions. Definition writing manuals are currently not very helpful regarding the contents of definitions.
My research aims at laying the groundwork for creating computer-assisted definition writing tools that can be used in any possible context and practice, by terminologists and ontologists, as well as domain experts such as scientists and lawyers. To help in definition authoring, such tools have to include widely usable definition templates. My work consists in creating such templates on the basis of the categories of the Basic Formal Ontology (BFO). Each template is composed of the characteristics of the different kinds of things that exist in the world: e.g., objects, processes, and qualities. To test to what extent these templates predict the contents of definitions, I describe the different parts of existing definitions from a large variety of domains written in different languages with the elements of the definition templates.
If the results show that a large part of the definition’s contents can be predicted with these templates, it will be possible to create definition writing tools that can be used in any context. Moreover, definition writing manuals can be completed with more precise principles regarding the contents of specialized definitions.
PROJECTS’ TITLES & FUNDING
My research on this topic was funded by the Swiss National Science Foundation from 2012 to 2015.
2014-2015: Consolidating the Ontological Analysis Framework for Future Applications: Follow-Up Project to “Reciprocal Enhancement of Terminological and Ontological Resources Through Definition Analysis”, Research project for the Advanced Postdoc.Mobility fellowship P300P1_154609 funded by the Swiss National Science Foundation (SNSF).
2013-2014: Reciprocal Enhancement of Terminological and Ontological Resources Through Definition Analysis, Research project for the Advanced Postdoc.Mobility fellowship P300P1_147821/1 funded by the Swiss National Science Foundation (SNSF).
2012-2013: Ontological Models for Feature Selection in Definition Writing for Terminological and Ontological Resources, Research project for the post-doctoral fellowship for prospective researchers PBGEP1-142281 funded by the Swiss National Science Foundation (SNSF).
2007-2008: Aspects conceptuels de la définition en terminologie, Research project for the doctoral fellowship for prospective researchers PBGE1-119326 funded by the Swiss National Science Foundation (SNSF).
- Definition Annotation Manual (Version 1)
- BFO-Definition-Models based on BFO 2.0 (available on demand)
- Poster summarizing the research project: