Domain-Independent Discourse Ontology for Utterance Interpretation and Generation in Human-Computer Dialogue

Abstract. This paper is concerned with the design, development and usage of a domain- and language-independent discourse ontol-ogy, useful for utterance interpretation and generation in service-oriented human-computer dialogue. Although the syntagm “dis-course ontology ” is not new, current work assimilates this term to rather domain-specific ontologies embedding discourse-related knowledge. In contrast with these approaches, a generic task-independent discourse ontology is proposed here; this knowledge base, structured in an object-oriented paradigm formalized in a first-order logic, can be used for rhetorical structure computation in human-computer dialogue. In this paper we will show how this ontol-ogy is structured, emphasizing the duality between physical structure (in the form of an XML database) and logical structure (enclosed in a set of axioms). Then, it will be shown how this ontology can be used for rhetorical structure updating in the process of (machine) utterance generation for human-computer dialogue.