Research Project: JurWordNet, law semantic lexicon
More info

JurWordNet is a domain lexicon; as in all technical fields, it is impossible to trace a borderline between common language and terminology: consequently, legal lexicon may not be considered as a subclass separated from generic resource. Overlapping occurs, both when the technical sense is stored in the Italian lexicon (ItalWordNet), and when the term is stored only in common meaning. The connection between the two resources is carried out manually, adopting pragmatic criteria relevant to the application task. The first simplification was to limit the connection to the sole nouns, as almost only nouns make the specific resource. A limited amount of adjectives is included, as they are polysemies compared to the nominalized use (private, public, …); verbs appear rarely in syntagms, and are connected to the noun that acts as subject/object.

Plug-in functions between legal resource and generic resource are of two kinds:

• specific meaning overlapping generic meaning (eq-lug-in) occurs when the sense defined in common language matches with the specific one. However, the definition (gloss) drawn from dictionaries or legal handbooks is more precise. In this case, through simultaneous access, all the lower knots of the trees are the specific ones, whereas the upper knots are those of common Italian. Likewise, ontological classification through Top Ontology, shared by all the resources of the EuroWordNet project and the connection with ILI.

• a specification of the legal meaning as opposed to the generic one (hypo-plug-in), which entails that the entire lower legal branch becomes a specific branch making up that of common Italian.

• based on the bottom-up building methodology, we abstracted legal terminology in general, and legally essential, categories. On the one side, this approach grants a satisfactory coverage of the dominion, and on the other, an assessment of the consistency of concept organization with the ontological one. The highest knots of JurWordNet are generally basic concept for the legislation world, and thus pertain to legal ontology (Core Legal Ontology). However, they are also language entities mapped with common language, and through it, can be matched to general ontological categorizing. This double matching is not inconsistent, but in line with the double perspective of using the database both as lexicon resource and as content description model.