Rose, C. P., Gaydos, A., Hall, B.S., Roque, A., & VanLehn, K. (2003). Overcoming the Knowledge Engineering Bottleneck for Understanding Student Language Input. Proceedings of AI in Education. Amsterdam: IOS Press.

In this paper we present Carmel-Tools, a new set of authoring tools for overcoming the knowledge engineering bottleneck for building tutorial dialogue systems that can accept natural language text input from students. Carmel-Tools provides the facilities for speeding up and simplifying the task of creating domain specific knowledge sources for sentence level language understanding. Carmel-Tools can build up domain specific lexical resources automatically from raw human tutoring corpora. It provides a GUI interface for authors to design their own first order propositional representation as well as facilities for annotating example sentences with their corresponding interpretation in this designed representational language. It then generalizes over these annotations and automatically generates all domain specific semantic knowledge sources required for interpreting novel input texts that are similar in content to those that have been annotated.

For a PDF full article version, click here (134KB).