Rhetorical Goals Surface Generator Functional Realizer Functional Description

In recent years, natural language generation research has begun to mature rapidly. It has witnessed both the appearance of sophisticated oo-the-shelf surface realizations systems and the development in the knowledge representation community of knowledge bases in a variety of domains. However, knowledge bases typically employ a logic formalism while the state-of-the-art surface realizer, Fuf, employs a uniication-based formalism to encode input expressions, grammatical knowledge, and lexical knowledge. As a result, bridging the gap between domain knowledge and linguistic knowledge has become a critical problem for natural language generation. We have designed a framework for functional realization that exhibits linguistic, domain, and task generality. It has been implemented in Fare, a robust functional realization system that translates logic-based representations found in knowledge bases to uniication-based representations that encode case assignments and phrasal constituent embeddings. Given a message speciication, Fare exploits lexical and grammatical annotations on knowledge base objects to construct functional descriptions, which are then converted to text by the Fuf surface generator. To study the generality of the approach, Fare was empirically evaluated in the context of the three diierent domains and two diierent applications. The results of these evaluations are encouraging.