Information Technology Reference
In-Depth Information
10. Examine whether example index table and domain name
π
exists, if not, return;
11. Delete corresponding table from base;
12. Delete domain name
π
from example index table;
13. IF
ρ
=4THEN /* adding examples to example base */
14. IF domain name
π
is not in the index table
call algorithm ANE to add new domain;
15. Check the appropriateness of record value and discard inappropriate record value;
store training set S0 into corresponding table in order;
16. IF
ρ
=5THEN /* deleting examples of example base */
17. IF domain name
π
is not in index table, then return;
18. Delete cases in table;
19. Extract(S
0
,
π
); /* generating case for typical case base */
20. Store cases;
7.8.7
Bias feature extracting algorithm
Features of candidate algorithms can help us to select optimum bias
automatically using optimal algorithm. These features are organized according to
the form of definition 7.19. The procedure will be activated when typical case
base or algorithm index table varies,.
Algorithm 7.10
Bias feature extracting algorithm
Input: classical example base CEB, register algorithm RA, example set ES, example
index table EI, algorithm index table AI, case base CB, operator OP
Output: modified CB
1. If op=1 // CB is empty
2. then { If EI = NULL || AI = NULL} return
3. For
i
=1 to maxRA
4. For
j
=1 to maxEI
5. For
k
=1 to Interval do
6. {using algorithm AI(
i
) to generate a classifier in example set EI(
j
)
7. and generate a data iterm according to definition 7.19
8. store data item into CB; }
9. Else if op=2 // adding a new algorithm
10. then { For
i
=1 to maxEI
11. For
j
=1 to Interval do
12. {using algorithm RA to generate a classifier in example set EI(j)
13. and generate a data term according to definition 7.19
14. store data into CB; }
15. Else if op=3 // adding a new example set
16. then{ For
i
=1 to maxRA
