Design of an Intelligent Support System for Scientific Databases
Abstract
This report describes progress made in designing an Intelligent Support System (ISS) for automatic data analysis and efficient data exploitation in numerical scientific databases, with an application to environmental databases used for ocean modeling and prediction. Improvements made to the Automatic Data Analysis System (ADAS) by combining the features of two machine learning techniques (i.e., data clustering and inductive learning by decision tree) to generate sets of production rules that efficiently describe the observational raw data contained in the scientific databases are explained. Data clustering allows the system to classify the raw data into clusters, which the system learns by induction to build the decision trees, from which are deduced the production rules. Also described is the design of a robust computational model of conversation (ENTRETIEN) for interactive natural language man-machine interfacing, based on case grammar theory. Such a system allows the user to naturally communicate with the system for efficient and effective information retrieval. ENTRETIEN conducts the conversation to recognize the user's intentions. To include the automatically generated production rules into the knowledge base of an expert system and to add the facilities of the natural language interface represent our final goal in building an intelligent support system for efficient exploitation of large scientific databases
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1994
- Accession Number
- ADA283043
Entities
People
- Frederick E. Petry
- Patrick Perrin
Organizations
- Mississippi State University