**5.3 Selection for** *pivots* **and** *targets*

As mentioned in 5.1, one objective of the algorithm is to minimize *slw*(*G<sup>R</sup> cls*, *φR*). Therefore, in *RDYs*, *pivots* should have maximum *LV* value (defined in table 1), because such a cluster may dominate *slw*(*G<sup>s</sup> cls*, *<sup>φ</sup>s*) and then *slw*(*Gs*+<sup>1</sup> *cls* , *φs*+1) after *s* + 1 th merging may became lower than *slw*(*G<sup>s</sup> cls*, *φs*). Our heuristic behined the algorithm is that this policy for selecting *pivots* can contribute to minimize *slw*(*G<sup>R</sup> cls*, *φR*). The same requirement holds to the selection of *targets*, i.e., *targets* should be the cluster which dominates *LV* value of *pivots*. In fig. 4 (b), *cls*2(3) having the maximum *LV* value in *RDY*<sup>2</sup> is selected. Then *n*<sup>2</sup> <sup>6</sup>, i.e., *cls*2(6) dominating *LV*2(3) is selected as *target*2. similarly in (c) *n*<sup>3</sup> <sup>5</sup>, i.e., *cls*3(5) dominating *LV*3(3) is selected as *target*3.
