##
**Specification and design of expert database systems.**
*(English)*
Zbl 0633.68110

Expert database systems, Proc. 1st Int. Workshop, Kiawah Island/S.C. 1984, 369-381 (1986).

[For the entire collection see Zbl 0598.00017.]

The article starts with a brief discussion of the role of expert systems (ES). ES should not give only precise answers to queries but should be also consultant to a user. Desired types of queries enabling ES to behave as a consultant are: “hypothetical queries”: what would be the consequence of X ?, “reason queries”: why would X happen ?, what would prevent occurrence of X ?, “candidates queries”: what are candidates for solution ?.

In the article a formal framework for specification of expert database systems (EDBS) is introduced. EDBS is recognized as dynamic object; static properties of such object are described by signature of modeling many-sorted algebra, dynamic properties are described by means of modal logic. The characteristic of similar approaches (Hoare-style logic, temporal logic) for describing dynamic properties of objects are briefly sketched. The paragraph dealing with related work is also valuable.

The design of EDBS is performed in two stages: (1) EDBS schema is specified; in this approach it means that a triple is defined: signature (i.e. a mapping assigning function type expression to every function and sort symbol to every variable), (possible empty) set of constraints on signature and (possible empty) set of inference rules on signature. (2) Signature is extended to update symbols, which are capable to express time dependent properties of EDBS.

For both above mentioned syntactical stages, the semantical ones follow. At first a set is associated with every sort symbol, and a function is associated with every function symbol. Further, in the second step a collection of update functions is defined; every update function acts on instances of EDBS and satisfies so called constraints.

The approach is well documented by examples, which demonstrate both database facilities of query language Varqa and inferentional features of ES-extensions. At the end the formulation of a hypothetical query in language Varqa is shown.

The article starts with a brief discussion of the role of expert systems (ES). ES should not give only precise answers to queries but should be also consultant to a user. Desired types of queries enabling ES to behave as a consultant are: “hypothetical queries”: what would be the consequence of X ?, “reason queries”: why would X happen ?, what would prevent occurrence of X ?, “candidates queries”: what are candidates for solution ?.

In the article a formal framework for specification of expert database systems (EDBS) is introduced. EDBS is recognized as dynamic object; static properties of such object are described by signature of modeling many-sorted algebra, dynamic properties are described by means of modal logic. The characteristic of similar approaches (Hoare-style logic, temporal logic) for describing dynamic properties of objects are briefly sketched. The paragraph dealing with related work is also valuable.

The design of EDBS is performed in two stages: (1) EDBS schema is specified; in this approach it means that a triple is defined: signature (i.e. a mapping assigning function type expression to every function and sort symbol to every variable), (possible empty) set of constraints on signature and (possible empty) set of inference rules on signature. (2) Signature is extended to update symbols, which are capable to express time dependent properties of EDBS.

For both above mentioned syntactical stages, the semantical ones follow. At first a set is associated with every sort symbol, and a function is associated with every function symbol. Further, in the second step a collection of update functions is defined; every update function acts on instances of EDBS and satisfies so called constraints.

The approach is well documented by examples, which demonstrate both database facilities of query language Varqa and inferentional features of ES-extensions. At the end the formulation of a hypothetical query in language Varqa is shown.

Reviewer: M.Benešovský

### MSC:

68P20 | Information storage and retrieval of data |

68T99 | Artificial intelligence |

03B45 | Modal logic (including the logic of norms) |