LingInfo: Design and Applications of a Model for the Integration of Linguistic Information in Ontologies

To allow for a direct connection of this linguistic information for terms with corresponding classes and properties in a domain ontology, we developed a lexicon model (LingInfo) that enables the definition of LingInfo instances (each of which represents a term) for each class or property. The LingInfo model is represented by use of a meta-class, which allows for the representation of LingInfo instances with each class, where each LingInfo instance represents the linguistic features of a term for a particular class. Applications of the LingInfo model are in information extraction, dialogue analysis, and knowledge acquisition from text, i.e. in knowledge base generation and ontology learning. 1. LingInfo: Motivation and Design To allow for automatic multilingual knowledge markup a richer representation is needed of the features of linguistic expressions (such as domain terms, their synonyms and multilingual variants) for ontology classes and properties. Currently, such information is mostly missing or represented in impoverished ways, leaving the semantic information in an ontology without a grounding to the human cognitive and linguistic domain. Linguistic information for terms that express ontology classes and/or properties consists of lexical and context features1, such as: • language-ID: ISO-based unique identifier for the language of each term • part-of-speech: representation of the part of speech of the head of the term • morphological and syntactic decomposition: representation of the morphological and syntactic structure (segments, head, modifiers) of a term • statistical and/or grammatical context model: representation of the linguistic context of a term in the form of N-grams, grammar rules or otherwise To allow for a direct connection of this linguistic information for terms with corresponding classes and properties in the domain ontology, we developed a lexicon model (LingInfo) that enables the definition of LingInfo instances (each of which represents a term) for each class or property. The LingInfo model is represented by use of a meta-class (ClassWithLingInfo) and meta1 Morphosyntactic and syntactic features may be based in future versions on the (ISO-TC37/SC4-MAF and ISOTC37/SC4SynAF) specifications. See also related documentation at the LIRICS project web site: http://lirics.loria.fr/documents.html property (PropertyWithLingInfo), which allow for the representation of LingInfo instances with each class, where each LingInfo instance represents the linguistic features (feat:lingInfo) of a term for a particular class. Figure 1 shows an overview of the model with example domain ontology classes and associated LingInfo instances. The domain ontology consists of the class o:FootballPlayer with subclasses o:Defender and o:Midfielder, each of which are instances of the meta-class feat:ClassWithLingInfo with the property feat:lingInfo.