// P is an n × n predicate similarity matrix, n is the number of predicates in ontologies
// δ is the threshold of silhouette width 
Input: P, δ
// a hierarchy with a set of clusters 
the jth cluster at ith level
Output:
1. i=1
2. repeat
3. // n is the number of input predicates, m is the number of predicates of cluster j at level i,
4. // optimal k (
) from m predicates of the cluster at level i (
) using 
function
5. 
//compute Neighborhood silhouette width
6. k = Optimal K(
) //fine the optimal k based on 
7. 
8. If (k > 1) then
9. 
10. for j = 1 to k
11. 
(// random mean for predicates in 
(cluster j at level i)
12. end
13. for each 
do
14. 
//wij is the degree, m is the fuzzifier
15. end
16. 
17. sw = 0
18. repeat
19. for each 
do
20. Update Cluster(
)
21. end
22. for each 
do
23. 
24. if 
then
25. 
26. 
27. 
28. end if
29. 
= Silhouette Width (
) //computesil houette width
30. while 
31. while 
and 