Database Reference
In-Depth Information
pected, the scalability of the border-based approaches is better when compared to
the exact approaches, since the runtime of the latter ones is primarily determined by
the time required by the BIP solver to solve the CSP. However, as shown in Figure
16.5, t
he runtime cost of solving the CSPs of the exact methodologies is worthwhile,
since the quality of the attained solutions is bound to be high.
Chess / Mushroom Datasets
20x10
15x10
10x10
5x10
Chess Dataset
20x7
15x7
10x7
5x7
20x10
15x10
Mushroom Dataset
10x10
5x10
20x5
15x5
10x5
5x5
0
1000
2000
3000
4000
5000
6000
Size of D
X
Fig. 16.7: Size of extension D
X
.
Chess / Mushroom Datasets
Chess Dataset
20x10
15x10
10x10
5x10
20x5
15x5
10x5
5x5
3x5
20x10
15x10
10x10
5x10
Mushroom Dataset
20x5
15x5
10x5
5x5
3x5
0
500
1000
1500
2000
2500
Constraints
Fig. 16.8: Number of constraints in the CSP.
