Susan Haller, Barbara Di Eugenio, and Michael Trolio
We present a method for aggregating information from an internal, machine representation and building a text structure that allows us to express aggregations in natural language. Features of the knowledge representation system, a semantic network, allow us to produce an initial aggregation based on domain information and the competing aggregate structures. In the final stages of realization, the network representation allows us to use low-level (below the level of a clause) information to perform linguistic aggregation. The test bed for this work is an interactive tutoring system on home heating systems.