Ontological Approach for Generating Natural Language Texts from Knowledge Base

Abstract

The computer systems are expert in dealing with structured data. However, when computer systems attempt to impart information to the end-users, generating natural language text that expresses structured data is a significant challenge. Currently graphical knowledge representations as a kind of forms to represent structured data are gradually becoming universal in computing. In this paper, we present a unified semantic model for generating fluent, multi-sentence, appropriate natural language text (e.g., Chinese language text) from knowledge base to the end-users. This article describes the development of semantic model for natural language generation, and the optional linguistic ontologies which may be used in the processing of generation. The main novelty is that it is possible to integrate different approaches and linguistic knowledge to generate natural language text from the structured data of the computer systems represented in graph form. For the ordinary end-users it will be an easier access to the information in the computer systems.