Using Application-Specific Ontologies to Generate Grammar Rules in English, Hindi, and Spanish for a Web Interface
Tutor / director / evaluatorGatius Vila, Marta
Document typeMaster thesis
Rights accessOpen Access
This thesis is concerned with the use of domain ontologies facilitating the generation of multilingual grammars, which furthermore can be integrated in the natural language understanding module of a communication system. In particular we work on grammars for supporting user queries when accessing the web in English, Hindi and Spanish in two scenarios: searching for a new medical specialist and looking for the information about cultural events in the city. Although there have been many works on communication systems supporting English and Spanish, this is not the case for Hindi language. For economical and cultural reasons there have not been many studies on the integration of Hindi language on communication systems. For this reason, our thesis also deals with the difficulty of working on a language for which not many studies have been done nor do existing resources exist. In order to facilitate the generation of linguistic resources for the three languages in different domains we propose a clear separation of the conceptual and linguistic knowledge, as well as a separation of general and domain-restricted knowledge bases, being conceptual knowledge. Conceptual knowledge is represented in ontologies and is reused across the three languages. Linguistic knowledge is language specific. General knowledge consists in general conceptual concepts represented in a general ontology and general linguistic knowledge can be represented as general grammars rules that can be reused across domains. For developing the grammar rules for each domain and language we use grammatical framework (GF), a powerful tool for writing multilingual grammars. One of the main advantages of this formalism is that favours a clear separation of conceptual and syntactic knowledge involved in a particular grammar: it represent the conceptual knowledge in a module called abstract grammar and the syntactic details in a separate but related module called concrete grammar. We define abstract grammar from ontologies and they are base to further developing the concrete grammar.