Thematic Domain Group: Lexical Semantics
NameLexical Semantics

TDG 13 for Lexical Semantics manages a profile that includes the set of data categories used for the representation of lexical semantic information in NLP lexicons. In order to allow coordination and interoperability, it becomes imperative for TDG 13 to establish synergies with other TDG s, Syntactic, Semantic, Semantic Role and Argument Structure, Ontology profiles, and, particularly, TDG 12, the Lexical Resource profile. The Lexical Mark-up Framework remains the privileged interface of TDG 13: its specific data categories are supposed to be used in combination with structural elements of the LMF – ISO 24613:2008. Lexical semantic data categories will supplement the structural lexical objects of the abstract LMF metamodel, thus becoming part of its definition and constituting the vocabulary used to express lexico-semantic information. This will allow lexicographers to implement concrete LMF lexicons and will help the LMF model to gain operability and usability. The activity in TDG 13 also aims at investigating and defining the constraints governing the relationships of these data categories with the metamodel and its extensions, mainly semantic and multilingual extensions.

Typical objects of investigation in TDG 13 are the data categories for the Lexical object SenseRelation, Synset Relation and Predicate Relation and their definitions. These data categories are gathered by looking at many best practices in semantic lexicon building: relations used in the framework of the Extended Qualia Structure to relate different senses, (is_part_of; used_for; created_by − ISOcat /isPartOf/; /usedFor/; /createdBy/); relations that link different synsets in lexicons of the WordNet family (inter-WN relations, that link synsets of a same WordNet, and intra-WordNet relations used to link WordNet in a multilingual fashion: has_synonym; has_eq_hyperonym – ISOcat /hasSynonym/; /hasEqHyperonym/); relations that are used to represent relations between Semantic Predicates (or Frames in a Frame Semantic Environment).

A domain information data category falls in the realm of TDG 13 as well, since they are used to represent the domain of use of a word meaning: medicine, biology, informatics, engineering. Contact points hold between the activities in TDG 13 and in TDG 3 Act.6, as concerns Semantic Roles used to specify a Semantic Argument with indication about its deep function: agent, patient. Relationships hold with TDG 6, as concerns the ontological nodes which are used to fill the MonolingualExternalRef and MultilingualExternalRef objects, having the specific purpose to align a meaning in the lexicon with a concept in a (shared) ontology. Ontological classes are surveyed in TDG 13 as possible descriptors to be assigned to semantic predicate’s arguments in order to impose selectional restrictions, as it happens in some well known lexicon practices. This allows predicting possible fillers of a semantic role relation among those senses labelled with the same node: human, animal, food…

A mailing list especially dedicated to TDG 13 activity is available and, at present, it counts 13 subscribers among the experts of the sector.

This profile is likely to include a number of Data Category Selections used in different LMF-compliant lexicon instantiations: the BioLexicon, NEDO Lexicon, the KYOTO WordNet-LMF grid.

Viewswitch to the TDG specific ISOcat view
(Note: this will reload the ISOcat UI in your browser and might require you to login again!)
ChairMonachini, Monica (ILC-CNR)