Verlyn Mark Johnson

Building a composite syntax for expert system shells

The authors examine the syntactic similarities and differences of five expert system shell production-rule languages. To help knowledge engineers manage information, they develop a composite production-rule syntax that provides a common language for defining production rules. The expert system shells are the: Integrated Reasoning Shell (TIRS); Expert System Environment (ESE), KnowledgeTool; Knowledge Engineering Environment (KEE); and Official Production System Version 5 (OPS5).

The San Francisco project: business process components and infrastructure

This paper provides an overview of IBM’s San Francisco Project. The goal of the project is to simplify the move to distributed object oriented applications. This is being accomplished by providing developers with an object oriented infrastructure and a starter set of object framework code for their applications. The project includes an infrastructure for distributed object applications, business objects that are common across multiple domains, and domain specific Business Process Components that contain a subset of the objects and business logic that are needed to build a commercial application. The Business Process Components are designed as business frameworks so that they can be changed and extended. Developers can learn to build object oriented applications by modifying the application code that is provided instead of having to design and build their applications from scratch.

Sharing and reusing rules-a feature comparison of five expert system shells

Compares the features we discovered while studying the production rules of five shells that reflect the products and features available on the market today: (1) The Integrated Reasoning Shell (TIRS); (2) Expert System Environment (ESE); (3) KnowledgeTool; (4) Knowledge Engineering Environment (KEE); and (5) Official Production System Version 5 (OPS5). These shells represent traditional approaches to production rule systems, as well as newer approaches such as using rules in a hybrid environment. One result of our research is an understanding of the features of these shells and how they affect the sharing and reuse of rule information. We have organised the material into tables on controllable system characteristics, system capabilities, production rule capabilities, and data manipulation capabilities. Each lists several features and summarizes whether or how each shell supports them.<<ETX>>