Natural Language Generation
Abstract
The report describes research and development of natural language generation and some parsing theory and technology in the Penman project of the Information Sciences Institute of Southern California. Penman, set of a natural language generation systems, provides computational technology for generating English sentences and paragraphs, starting with input specifications of a non- linguistic kind. Its associated information resources, such as high-level concept models, domain models, lexicons, etc., are also employed for other uses, including information retrieval and machine translation. The following five areas of research and development were addressed during the funding period being reported on in this document: (1) Sentence generation: continued expansion and distribution of the Penman sentence generation system; (2) Multi sentence text planning; development and testing of new text structure planning technology; (3) Parsing: construction of a prototype parser using the Penman system's grammar and semantic models; (4) Information retrieval: development work toward the eventual collaboration with two other sites in the joint construction of a new MT system.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 31, 1991
- Accession Number
- ADA264929
Entities
People
- Eduard H. Hovy
Organizations
- University of Southern California