Ontology-Based Natural Language Texts Generation from Knowledge Base

Abstract

AbstractIn the knowledge-based intelligent system, the natural language interface is responsible for implementing communication between end-users and intelligent system using natural language. However the internal information of intelligent system is represented in a certain kind of specific format (knowledge base). When intelligent system need to exchange information with the end-users, it is necessary for natural language interface to obtain the ability to convert fragment of the knowledge base into natural language text. Due to the diversity and complexity of natural language, generating fluent, appropriate natural language text is still a significant challenge. By comparing other methods converting structured data into natural language text, this article describes the characteristics of fragment of knowledge base and the difficulties in natural language generation. The article proposed a unified semantic model to development of natural language generation component of natural language interface. The development of natural language generation component requires a combination of various types of knowledge about linguistics and various problem solving models oriented on natural language generation. In this article the main novelty is that in the unified semantic model the various approaches and linguistic knowledge about natural language generation can be integrated to generate natural language text from the fragment of knowledge base represented in graphical form.Keywordsnatural language generationOntologyKnowledge baseNatural language interfaceKnowledge-based system