INTEGRATING LANGUAGE AND MEANING IN STRUCTURED INHERITANCE NETWORKS

Abstract

Natural language systems must map between language and meaning. The mapping process is central to language analysis—producing an internal representation of an input text, and language generation—producing language to express an underlying knowledge representation. This language processing issue suggests using a representation that explicitly associates concepts with linguistic structures. Extending structured inheritance networks to include these associations helps in building efficient natural language processing systems that can perform both analysis and generation, and extend easily to new domains. The integration of language and meaning requires a representation that provides a general framework for mapping but adequately constrains the mapping process.