Program 4. Sub sum Check
parameter a node N(X), a TU-Table TU and the min_utility threshold. The system
initially performs Sub sum Check on X as displayed in Program 3. This check
confirms if there exists an item from PREV-SET(X) such that g(X) ? g(a). In the event that there exists such an item, it implies that X is
incorporated into a closed itemsets that has already been found and supersets
of X don not need to be investigated (see 14 for a total justification).
Something else, the next stage is to process the closure XC = C(X) of X. This
is performed by the method Compute Closure (N(X), POST-SET(X)) appeared in
Program 4 14. At that point the estimated utility of XC is ascertained. In
the event that the condition is not less
than min_utility, XC is considered as a candidate for Phase II and it is
yielded with its estimated utility value Est U(XC). Note that CHUD does not
keep up any discovered applicant in memory. Rather, when a candidate item sets
is discovered, it is outputted to disk. After this, a hub N(XC) is made and the
technique Explore is called for discovering candidates that are supersets of XC.
Strategy 1. Considering only just encouraging items. The
first procedure that we have incorporated in CHUD is to consider only promising
items for creating candidates and to remove the utilities of unpromising items
from the GTU table. It is connected in line 2 and 3 of the Main procedure.
Rationale. It was appeared in 14 that unpromising items cannot be part of a
HUI and that the utility of unpromising items can be overlooked in the count of
the estimated utility of item sets when searching for high utility item sets.
Strategy 2. Discarding itemsets having an expected
utility lower than min_utility. The second procedure in CHUD is to discard the itemsets XC to such an extent that Est
U(XC) ? min_utility. This methodology is incorporated in line 3 of the CHUD
Phase-I strategy. Rationale. It was shown in Section 2 that an itemsets that is
not a HWTUI is not a high utility itemsets and in addition the greater part of
its supersets (see Property 1 and Definition 4, 8 and 9). Since DCI-Closed
discovers candidates recursively by considering supersets of candidates,
disposing of an itemsets such that EstU(XC)