Natural language generation systems : a comparative study

The development of natural language generation systems has been one of the most considerable subjects of research in Artificial Intelligence during the last decade. This paper investigates several important research projects in natural language generation developed up to date. Its aim is to provide not only a complete presentation of some of the most significant work in the fields of AI and Computational Linguistics, but also a broad, comparative study of some generation systems. For this purpose, we identify particular techniques in each system which contribute strongly to the quality of the resulting natural language text. Comparison of their individual skills then reveals their performance capabilities and limitations as well as common points, differences and possible advantegeous combinations of these techniques that could lead to even more effective systems.Preprin