Lexical Conventions
Anita Liang
in collaboration with GI-FAOTERM, Terminology, GICM, FAO 2005
|
Lexical Conventions
Anita Liang
|
|
|
|
|
|
|
Lexical conventions create consistent formats for enhancing the effectiveness and efficiency of terminology-related activities. They improve the interpretability of lexical data by humans, serving as a form of documentation (e.g., allowing quick identification of names), and facilitate machine processing (e.g., enabling programmatic manipulations of terms).
The current state of AGROVOC is the immediate motivation for the development of these conventions as it suffers from discontinuities in its development, including a history of changing managers, differences in the conventions (if any) used in the maintenance and extension of the thesaurus during that time, and the application of standards (e.g. ISO 5964 for multilingual thesauri) that in and of themselves do not foster consistent formats. The result is inconsistencies throughout the thesaurus, e.g., in the use of case and singular/plural forms, both within and across languages, which have, in turn, led to other unwanted consequences, e.g., inaccurate translations and classifications.
Apart from the goal of "cleaning up" the terms in the thesaurus, there is also the need to establish lexical conventions for ontology development in anticipation of the medium-term objective to convert AGROVOC into an ontology and of future projects for developing other ontologies within the auspices of the AOS project. The conventions specify consistent labelling and identification of concepts, relations, and attributes while adhering to the rules for legal XML names
Thus, two sets of conventions are specified: one applicable to vocabularies currently being maintained at FAO, that are used mainly (though not exclusively) by humans and that can be applied immediately, and the other to machine-processible semantic structures such as ontologies.
For more information, cf. Soergel et al. (2004), Soergel Consultant's report, "Top Quadrant's Guidelines for Ontology Modeling," "Propositions [sic] of Conventions for RDF," etc.