Inductive Learning - Advanced Artificial Intelligence - page 296

1. Call preprocessing algorithm PPD;
2. Call seek optimum algorithm SM to get optimum algorithm Γ ;
3. Search algorithm Γ in appointed path;
4. If algorithm exists,
5. then Call Γ ;
6. Generate decision tree and/or production rule set;
7. Store learning result into target knowledge base;
8. If algorithm does not exist,
9. then let ρ =3, call case base maintain algorithm CBM to delete Γ from case base
and corresponding data structure;
10. go to step 2;
11. Fill out the journal for BSDT.
7.8.6 Typical case base maintain algorithm TCBM
Algorithm TCBM realizes various operation to typical case base, such as adding
new example set, example set version updating, deleting example set, deleting
example set and adding or deleting examples of example base, etc. Only when
add or delete new example or modify existing example subset, the algorithm can
be activated. Because of employing hierarchical activating mode, after TCBM
algorithm modify classical case base, it would automatically call each algorithm
in algorithm index table to execute again to construct case base in accordance
with the content of classical case base. Case base and modification of related data
structure are selected to influence optimum algorithm directly.
Algorithm 7.9 Typical case base maintain algorithm TCBM
Input: Primitive case base, case index table, case set S 0 , domain name π , maintain
operator ρ
Output: Modified case base, case index table
1. IF ρ <1 and ρ >5 THEN return;
2. IF ρ =1THEN /* adding new case set */
Call ANE algorithm to add new case;
3.IF ρ =2THEN /* typical case base version updating */
4. examine whether example index table and domain name π exists;
5. IF exists
6. delete corresponding table from case base;
7. delete related items in case index table;
8. call algorithm ANE to introduce new case set;
9. IF ρ =3THEN /* deleting case set */