AIS: a foundation for practical artificially intelligent systems

The design of the Artificial Intelligence Substrate (AIS) is presented. It is a Lisp-based knowledge representation language (KRL) to support large-scale distributed artificial intelligence (DAI) applications. AIS uses a two-level data model. One level acts as a permanent repository of information and the other contains a partial working set of it. The basic data model stores information as entities and relational tuples. Demons, which are procedures that respond to pattern-matched transactions, are used to maintain consistency of the database and to automate standard activities. On top of the entity-relationship data model an object-oriented scheme has been implemented. A forward- and backward-chained rule-based system also has been developed in terms of the object-oriented system. Each of these knowledge representation schemes have associated optimization strategies. Finally, an automatic classification algorithm is available so that domain models can be constructed from exemplars