Information Systems Reengineering, Integration and Normalization - Information Systems Reengineering, Integration and Normalization - page 23

reverse engineering. To support this process we need an ES shell. This consists of
an inference engine, a factbase, and one or more rule bases. The database schema is
automatically converted by a preprocessor into a factbase for the ES. Each record
name is translated into a fact statement. For example, derive facts from the given
DDL. (Fong and Ho 1993 ).

DDL FACTS

RECORD department department is a record

dept PIC 999 dept is contained in department

dept-name PIC CHAR(30) dept-name is contained in

department

RECORD instructor instructor is a record

name PIC CHAR(30) name is contained in instructor

instr-addr CHAR(50) instr-addr is contained in instructor

RECORD section section is a record

section-name section-name is contained in section

SETdept-instr dept-instr is a set

OWNER dept dept is contained in dept-instr

MEMBER instructor dept owns instructor

SET instr-sect inst-sect is a set

OWNER instr instr is contained in instr-sect

MEMBER section instr owns section

section owns none

The following backward rule transforms the records into entities and 'R' represents
variables to be instantiated:

'R' is a entity /* known facts if the condition is met */

If

'R' is a record /* the condition */

The ES shell provides a mechanism to obtain facts from users in the form of “ask-
able facts,” such as 'E' identified fully? When 'E' is bound to department, for ex-
ample, will generate
Is the statement: department identified fully, true? Please enter (Y)es, (N)o or
(W)hy.
Typing “why” will generate an explanation of why the system asked the ques-
tion, by showing the rules that may help the user to respond better. If the answer is
“yes,” the entity is tagged as fully internally identified and the premise succeeds. If
the answer is “No,” this premise fails. For the conclusion to fire, the premises must
succeed, otherwise, the system will try the next rule.