Summary of important information

1. Mining Maximal Generalized Frequent Geographic Patterns with
Knowledge Constraints

2. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern Mining
4. Mining MGFGP with Knowledge Constraint
5. Conclusions and Future Work

3. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
Increasing large number of frequent itemset and rules in
Frequent Pattern Mining ( FPM ) will be overcome by
proposing massive sets of algorithms for transactional
database without initial information in lowering the non-
interesting pattern. This is caused by the natural
geographic data, called a priori.
Geographic data includes visible pattern that highly
connected to the geographic dependent and does not
impact the discovery of novel and useful knowledge.
Strong geographic domain such as a(GasStation)->
intersect(Street)100% will not be considered by some
user, but to the a(GasStation) and
intersects(WaterrResource->pollution=high(40%) instead.

4. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
The objective of the study is to reduce the well-known
patterns and redundant frequent sets by omitting the
dependences in a first step and all redundant frequent sets
in a second step, computing maximal generalized frequent
geographic patterns (MGFGP). Section 2 will study the
related works and the main contribution of this paper.
Section 3 describes the problem of mining frequent
geographic patterns with well-known dependences.
Section 4 presents the algorithm MG-FGP and shows
experiments performed over real geographic databases.
Section 5 concludes the paper and gives directions of
future work

5. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
First approach used is quantitative reasoning that
calculate the distance relationship on frequent set
generation. The algorithm used includes direct geographic
attributes, points restricted, only calculate qualitative
spatial relationship on the frequent set and ignore the
significance of non-spatial attributes of geographic data.
Second approach [3][4]is qualitative reasoning that focus
on both distance and topological relationship of
geographic object type and a set of relevant spatial feature
types on geographic primitive. Spatial
relationship is implemented in first step while the other is
calculated in another process. Both approaches have not
focused on interesting geographic aspects to be
considered in FPM.

6. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
Though well-known geographic pattern is lowered in [5]
and well-known pattern is reduced in [6], large number of
frequent sets are still created. Subsequently, the study will
implement the close frequent techniques set [7][8]in the
process of mining the frequent geographic patterns
excluding an inoperable frequent set. Based on the result
obtained at closed frequent set when they are applied to
the geographic domain, the study then proposes the
absence of well-known dependence to produce non-
inoperable frequent set.

7. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
The general problem in mining FGP with well-known
dependence are divided into two, the one happens within
geographic dependences in FPM and the other that
happens in geographic dependences of closed FPM. In the
first problem, the study will elaborate the details definition
of the transactional FPM such as row ( target feature type )
and columns that is called as predicate. The study later
explains the set F : = {f1 , f2, ...,f k, ..., fn} which is a set of
non-spatial attributes and special predicate.

8. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
The two main problems in geographic FPM are as follows:
a) extract spatial predicates , where a spatial
predicate is a spatial relationship (e.g. distance)between
the target feature type and a set of relevant
feature types; and (b) find all frequent predicates, that
means a set of predicates is frequent if its support is at
least equal to minimum support.

9. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
Based on the [1], every subset in Z depends on its original
predicate Z size. The image in figure 1(a) illustrates the data
set that include 6 tuples and 5 predicates such as port,
school, water resource, hospital and treated water
network).It is clearly indicated that there has been
geographical dependence between port (A) and water
resource (W), where cities with port will also contain water
resources. This also indicates that A or W correlates to the
C,D or T and they cannot be eliminated.

10. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
On contrary, the figure 2 indicates that when A and W are
replicated into many larger frequent sets,the information
in [5] and [6] would remain the same even if A and W
eliminated. Subsequently, there will be many inoperable
frequent sets ( eg A,C},{A,T},{C,T},{T,W) occurs where will
be settled by closed frequent set. In the closed FPM, it
indicates that based from the [7] and [8], the closure
operator will proceed the definition of all closed frequent
itemset and maintain the minimum amount of non-
operable frequent sets.

11. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
The figure 3 illustrates the criteria of frequent item set
which symbolized by minimum support received and also
because the set of transaction occurs in the data set is less
than {A,D,W}. The figure also shows that to achieve the
frequent geographic pattern, closed frequent set
technique is not applicable to geographic data. To solve
this problem, the study proposes the MGFGP – Maximal
Generalized
Frequent Geographic Patterns without well-known
geographic dependences, where in figure 4, it clearly states
that transactions 135 runs without well-known
dependences ({T,W}, {A,T}, {A,D,T}, and {D,T,W}.

12. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
Later in the figure 4, it shows that within one frequent set
is closed, two frequent sets are proved to be maximal
excluding its well-known geographic dependences. Though
the number of frequent set is reduced following the
elimination of the geographic dependences as well as
redundant frequent, the information resulted is still
complete and the quality is still ruined.

13. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
The figure 6 demonstrates the method of generating maximal
generalized frequent geographic patterns without well-known
dependences by seudo-code of the algorithm MG-FGP. The
details are as follows; MG-FGP diminish all predicate sets with
geographic dependences and while in apriori, the MG-FGP will
passes several times over the data set which later used to
determine the k-predicate sets. Later, the k as the number of the
current pass, the large sets Lk-1 in the previous pass (k -1) are
grouped into sets Ck with k elements working as the candidate
sets.

14. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
A set will be considered frequent when the computed candidate
resulting similar or exceeding minimum support and will
experience the process until the number of frequent is 0. The
second pass will generate candidates with 2 predicates and
permit the existence of geographic object and appear in the
spatial association rules.
When the MG-FGP proceed the generalization, the Gk will be
removed form G when transactions (tidset) in which Gk appears
is the same set where Gk+1 appears, and the set Gk ⊂Gk+1, then
we can say that Gk is inoperable. Subsequently, after the L and G
measured, spatial association rules will be created to produce
non-redundant association rules.
The formula is tested in real-life condition where the
representative locations are : waterbodies, hospital, street, slum,
gas stations and etc. The study then deletes the dependency
between bridge and water as well as bridge and water body and
later perform 5% and 10% minimum report.

15. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
Figure 7 indicates the result after the number of frequent sets is
reduced is in average of 87% . However, geographic dependence
existed when frequent sets generated for minsup 5%, 10%, and
15%. Closed frequent set approachterminate a large number of
frequent sets, in which 50% of them are well known geographic
pattern. The MG-FGP is found to be more effective than closed
frequent as seen in the figures 8 where the frequent sets the
closed frequent set approach, which both do not eliminate
geographic dependences

16. Summary
1. Introduction
2. Related Works and Contribution
3.The General Problem of Mining FGP
with Well Known Dependence
3.1 Geographic Dependences in Frequent Pattern Mining
3.2 Geographic Dependences in Closed Frequent Pattern
Mining
4. Mining MGFGP with Knowledge
Constraint
5. Conclusions and Future Work
To conclude the study, it is well known to mining a large number
of patterns and based on the real-life application as the closed
frequent set approach produces many closed frequent sets
containing well known dependences. Method used in eliminating
well known geographic pattern is by omitting geographic
dependences in a single step to avoid the redundant association
through eliminating and and maintaining inoperable frequent
set.