GEA: A Complete, Modular System for Generating Evaluative Arguments

摘要

This paper presents a system for generating user tailored evaluative arguments, known as the Generator of Evaluative Arguments (GEA). GEA design is based on a pipelined architecture commonly used in natural language generation. After an overview description of GEA main components, we focus on how GEA performs the microplanning tasks. Details are provided by examining the generation of a sample argument.