![](bg3.png)
linguistic expression “voltage threshold relay” is not
in usage, but can be defined as a normalised term.
Although we can extract some useful information
from texts, ontology cannot be built directly from
them since we need ontology for understanding text
(understanding text requires extra-linguistic
knowledge which by definition is not included in the
corpus).
This is why we have introduced the new
paradigm of ontoterminology (Roche 2007) to take
into account these two different activities –
conceptualisation and writing text – and to focus on
conceptualisation. The main goal of terminology is
first to understand and conceptualise the world and
then to name it. Ontoterminology allows building a
new kind of terminology in which the concept plays
a central role. An ontoterminology is a terminology
whose terms, either of usage or normalised, are
related to concepts defined in a formal ontology.
This makes it possible to manage the linguistic and
conceptual dimensions of terminology and provide
two kinds of definition: the first formally defines the
concept whereas the second explains the term and its
usage from a linguistic point of view.
2.2 Term and Concept
Concepts in ontoterminology exist in their own
right. Thus, ontoterminology manages terms as well
as concepts; both are entries in this new kind of
terminology. It also means that term and concept
definitions are separate but connected since the
meaning of a term is related to a concept. In the
example below (see figure 3), these definitions
appear in two different cards, one for the concept
and another for the term.
Ontoterminology enables focusing on the
conceptual and linguistic dimensions of
terminology. Terms and concepts belong to different
and non-isomorphic semiotic systems. In order to
show such a difference, terms, as linguistic
expressions, are written between quotation marks
e.g. “turbine”, while concepts, as entities of a formal
system, are written between chevrons and start with
an upper case e.g. < Hydraulic turbine>.
If ontoterminology enables normalisation of
language, unlike classical terminology it also
enables preserving the diversity of language between
different communities of practice since they share
the same domain conceptualisation. In point of fact,
two different terms can denote the same concept
whose name should be written so that we understand
the right place of the concept in the ontology. Such
concept names define normalised terms which
cannot be used in text (e.g. because they are too
long) but are necessary for term meaning and
understanding. For example “voltage relay” in
English and “relais de tension” in French denote the
same concept of <Voltage threshold relay>.
2.3 Conceptual Structure
The conceptual relationships are used for structuring
entries. In figure 3 the concepts are listed in
alphabetical order combined with either the “is-a” or
the “part-of” relationship. These conceptual
relationships are also used for building the lexical
structure which is automatically updated each time
the conceptualisation is modified.
Words and linguistic relationships are no longer
the only means to access information in
terminology. Associating information to concepts,
e.g term definitions, documents, returns on
experience, etc., amounts to classifying expert
knowledge in the terminology.
It is also possible to define new paradigms of
navigation based on the domain ontology. Ontology
can be viewed as a conceptual map (Tricot et al.
2005) in which the experts navigate along the “is-a”
and “part-of” relationships in order to access
information connected to concepts (figures 3, 4 and
5).
Schemas play a key role in technical domains.
From the conceptual point of view, they represent
one of the most important references. Experts agree
on this kind of independent natural language
knowledge, easier to understand and more
consensual than texts. They refer to schemas every
time a communication problem occurs or when an
explanation is required. A schema describes a
physical entity and the parts which make up it. Each
of these parts is also described by its own schema.
Entities and components are modelled by concepts
linked by the part-of relationship. These concepts
create a network of part-of linked concepts which
allows users to browse from a schema describing the
current concept to a more detailed or global schema
associated to one of its part-of concepts. Just as
hypertext has defined a new method of corpus
navigation using textual links, hyper schema defines
a new method of knowledge base navigation
attached to the domain ontology using conceptual
links (see figures 4 and 5).
ONTOTERMINOLOGY - A New Paradigm for Terminology
323