Building a Parallel Spatio-Temporal Data-Text Corpus for Summary Generation
Ross Turner, Somayajulu Gowri Sripada, Ehud Reiter, Ian P. Davy · 2008
We describe a corpus of naturally occurring road ice weather forecasts and the associated weather prediction data they are based upon. We also show how observations from an analysis of this corpus have been applied to build a prototype Natural Language Generation (NLG) system for producing road ice forecasts. While this corpus occurs in a narrow domain, it has much wider applicability due to the nature of its spatial descriptions, whose primary communicative goal is to describe the interaction between meteorological parameters and geographic features. 1.