A Natural Language Database Interface For SQL-Tutor
Seymour Knowles
Department of Computer Science
University of Canterbury
Abstract
An investigation into integrating a database Natural Language Processing (NLP) component into the SQL-Tutor Intelligent Tutoring System (ITS) is presented. Tailor-made NLP systems created by a programmer, and NLP systems created automatically by a general database NLP system, are considered with respect to the requirements of SQL-Tutor. A tailor-made system is created for the MOVIES database using a semantic grammar, and its strengths and weaknesses are demonstrated. Three levels of information are identified in the tailor-made system; database independent information, database structure and database semantics. These levels are used to assess a commercial system for automatic database NLP generation. Some weaknesses are identified in this system and a proposal of an `authoring tool' to remedy these weaknesses is presented.
