A natural language for implementing algebraically expert systems
Antonio Hernando, Roberto Maestre Martínez, Jesús Bobadilla, Eugenio
Roanes-Lozano
Key words: Expert Systems, Groebner basis, Knowledge representation
According to previous mathematical results, expert systems can be quickly
and easily implemented on specialized mathematical software (`Computer
Algebra Systems'). These mathematical results provide two equivalent
different approaches to implement algebraically expert systems: one based on
propositional Boolean logic and another based on the `concept-attribute-value'
representation paradigm. Both approaches are equivalent in the sense that
every expert system implemented through one algebraic approach can be also
implemented on the other algebraic approach.
However, the performance of these algebraic approaches depend on the specific
expert system, and therefore, the expert system must be implemented in both
algebraic approaches in order to compare their performances. In this paper we
have designed a natural language to implement algebraically expert systems
which provides these three main advantages: we can implement directly any
expert system in both algebraic approach since our language is independent
of the algebraic approach used; we can use any computer algebra system
since our language is independent of the computer algebra system used;
and we do not need to be acquainted with specialized abstract mathematical
concepts like Groebner basis or normal forms since our language is natural
and does not make use of these concepts.