Knowledge-base Systems (KBS)-Case-based Designer's Assistant 

Goal: The primary goal of the KBS project was to effect revolutionary advances in the state of the art in machine intelligence applied to the design domain. The proposed project produced a prototype intelligent designer's assistant tool The prototype embodied innovative features to enhance designer productivity by automating in inclusion of design history, design knowledge, and features-based geometric objects.

Results: The KBS project study tool demonstrated revolutionary advances in machine/learning by producing innovations in the following areas: Case-based Reasoning (CBR), neural network processing, expectations-based user modeling, and explanation-based processing. These novel techniques were combined with the following existing methods: feature-based reasoning, rule-base reasoning, and object-oriented programming.

The important result will direct SOA toward the proliferation of intelligent assistants which learn from their users in any domain of design. The KBS project interactively suggested preliminary design changes based on learned knowledge related to parts produceability, inspectability and cost. The designs were within a family of parts from the advanced composites domain.

An innovative method of using expectations processing to track designer goals and actions were developed. A unique scheme using explanation-based processing to prompt the designer for decision rationale resulting in a novel design knowledge capture capability.

Measurements: The project was within schedule and budget, and met all milestone award requirements of the customer. The global innovative claim is that the KBS project was developed in such a way that specific design knowledge was removable from the system. This would enable the prototype to serve as a shell, or tool, increasing its ease and benefit of application to other domains.

The project results were feed into the long range goal of eventually implementing intelligent assistants in other design domains. By learning from designers, these assistants will someday automate the entire design process.