Evaluation of Proposed Algorithm with Preceding GMT for Fraudulence Diagnosis

Navneet Kr. Kashyap

Research scholar, Department of Information Technology, Govind Ballabh Pant University of Agriculture and Technology, Pantnagar-263145, Uttarakhand, India

Corresponding Author Email: er.navneetkashyap@gmail.com

DOI : http://dx.doi.org/10.13005/ojcst/9.02.01

ABSTRACT:

Formerly existing graph mining algorithms regularly accept that database is generally static. To defeat that we proposed another algorithm which manages extensive database including the components which catches the properties of the graph in a couple of parameters and check the relationship among them in both left and additionally right course, in this way embracing DFS and in addition BFS approach. It furthermore discovers the subgraph by traversing the graph and removing the planned routine. The proposed calculation is utilized for identification of wrongdoing as a part of BANK & Financial organization by catching the properties and distinguishing the relationship and affiliations that may exist between the individual required in that wrongdoing which keep a few violations that may happen in future. We have utilized the Neo-ECLIPSE for the execution of proposed calculation and Neo4j is the graph database utilized for evaluation.

On the off chance that a man  endeavoring to confer fraud or engage in some kind of illicit movement, they will endeavor to pass on their activities as near authentic activities as could reasonably be expected. Here in this paper, we are giving the data that a man who is in beginning the phase of the fraud, what co-related wrongdoings or illicit exercises he can do in future. The future exercises that can be performed by the individual can be ceased by demonstrating the associations with the entries saved in the database.

KEYWORDS: Part Miner; gSPAN; gIndex; Graph database; Traversing

Copy the following to cite this article: Kashyap N.K . Evaluation of Proposed Algorithm with Preceding GMT for Fraudulence Diagnosis. Orient.J. Comp. Sci. and Technol;9(2)

Copy the following to cite this URL: Kashyap N.K . Evaluation of Proposed Algorithm with Preceding GMT for Fraudulence Diagnosis. Orient.J. Comp. Sci. and Technol;9(2). Available from: http://www.computerscijournal.org/?p=3661

Introduction

These days the measure of information is expanding step by step, so appropriately the longing for information mining is likewise developing. The substantial database must be looked to locate the fascinating properties of the graph and to build up a relationship along with them. It is gainful to demonstrate the complex data with the assistance of graph in which data is stored in nodes and edges speak to the relationship among the nodes [1][2]. Subsequently having a Graph database defeats the important of a relational database and helps in finding the supergraph, subgraph, basic graph and connection in between different graphs. This graph based information mining has turned out to be increasingly well-known in the most recent couple of years [3]. Graph mining is the utilization of most essential structure of graph to acquire regular patterns of data. It has board scope of uses [4].

This graph-based data mining has turned out to be increasingly famous in the most recent couple of years. Graph mining is the utilization of most essential structure of graph to get regular patterns of data. It has board scope of applications. This procedure can be utilized to discover the possibility of persons doing wrongdoing in the organization through the web or by using any other way .Some relevant researchers of individuals required in digital wrongdoing were concentrated on to get the characteristics, for earning, persons required in wrongdoing, whether they are taught or not, style of wrongdoing, acquiring from the specific risk [5]. These feature lead to the development of graph database and algorithm happens to be proposed for traversing the graph in both headings left and in addition right and build up relationship among various nodes   which assist creates a sub graph as per the request.

Neo4j is the graph database utilized for evaluation as the recovery times of graph database are not exactly social database as it takes a look at records, it doesn’t check the whole gathering to discover the nodes that met the inquiry criteria [14][7]. Analysis report from this execution will likewise be useful in arranging the prevention concerning a number of offenses. The rest of this paper is sorted out as takes after.

Part 2: presents the issue proclamation of graph based information mining and existing calculations;

Part 3: illustrates our proposed calculation utilized for traversing the graph database;

Part 4: present similar investigation of our proposed system with other existing procedure;

Part 5: conclude Conclusion and future expansion.

Overview Of Existing Algorithm

Part Miner Algorithm

Every graph in the database is divided into littler subgraphs. Part Miner can viably diminish the quantity of candidate graphs by examining the total data of the units. This has prompted a considerable measure of cost investment funds saving. Part Miner is successful and adaptable in discovering subgraphs [6].

 Algorithm Graph Part 

Input: G, the graph

Output: G1, G2, the two subgraphs of G

1: V = {vertices sorted according to

Their update    frequency};

2: V*= Ф;

3: w (V*) = −∞

4: for (i = 0; i < |V |/2; i++) {

5: Vi = Ф;

6: call DFSScan(V, i, Vi);

7: Compute w(Vi);

8: if (w (Vi) > w(V∗ )) {

9: w (V*) = w (Vi);

10: V*= Vi;

11: }

12: }

13: G1 = {eij = (vi, vj )|vi ∈  V*, vj ∈  V*}

∪{eij =   (vi, vj )|vi ∈  V*, vj /∈ V*}

14: G2 = {eij = (vi, vj )|vi /∈ V*, vj /∈ V*}                  ∪ {eij = (vi, vj )|vi ∈  V*, vj /∈ V*} 

Procedure DFSScan(V, i, Vi)

15: stack = Ф,m= 0;

16: stack.push(vi);

17: while(stack ≠Ф∧  m ≤ |V |/2){

18: v = stack.pop();

19: Vi = Vi ∪ {v};

20: m++;

21: choose the neighbor vertex vh,

s.t. vh.visited = 0, and ∀ vs,    

Vs.visited = 0 ∧  (v, vs) ∈  E, vs.ufreq < vh.ufreq;

 22: stack.push(vh);

23 :} 

Dividing graph database into units

Procedure DBPartition(D, k)

D, graph database;

K: number of units

1: D0,0 = D;

2: i = 1;

 3: l = log2k;

 4: while (i ≤ l) {

5: for (j = 0; j < 2i−1; j++)

6: DivideDBPart(Di-1,j,Di-2,j,Di,2j+1);

 7: i++;

8 :}

9: for (j = 0; j < k – 2l; j ++)

10: DivideDBPart(Di-1,j , U2j , U2j+1);

Function DivideDBPart(Ds, D1,0, D1,1)

1: D1, 1 =Ф;

2: D1, 1 = Ф;

3: for each graph G ∈  Ds {

4: G1, G2 = calling GraphPart(G);

5: D1, 0= D1, 0 ∪ {G1};

6: D1, 1 = D1, 1 ∪ {G2}

gSpan Algorithm

Graph-Based Substructure Pattern Mining that introduced the gSpan algorithm which usually finds out regular substructures without having candidate production. gSpan develops a new lexicographic arrangement among the graphs and routes every graph to an exclusive minimum DFS code as the canonical label. Dependent upon this lexicographic order, gSpan explores the depth-first search approach to exploit regularly connected subgraphs effectively[7][8]. So, gSpan outperforms FSG by the order of degree as well as is suitable to exploit huge regular subgraphs in a larger graph arranged with lower minimal help.

 GraphSetProjection(D,S).

 1: arrange the labels in D by their regularity;

2: eliminate occasional vertices and edges;

3: relabel the leftover vertices and edges;

4: S1← all regular 1-edge graphs in D  ;

5: sort S1 in DFS lexicographic order;

6: S ←S1

7: for every edge e € S1 do

8: initialize s alongside e, set S. D

by graph which includes e

9: SubgraphMining( D,S,s);

10: .D←D-e

11: if │D│< min Sup

12: break;

 Subprocedure 1 SubgraphMining(D,S,s)

1: if  s ≠ min(S)

3: S←S U 

4: specify s in every graph in D

and count its children;

5: for each c, c is s’ child do

6: if support (C) > min Sup

7: s←c

8: SubgraphMining(D,S,s_);

gIndex Algorithm

Assorted out from the established route-based techniques, this strategy, known as gIndex, will make use of regular substructure as the fundamental categorization or indexing property [9].  Frequent  substructures  tend to be  appropriate  candidates considering  that they  search  the  internal  attributes  of  the  information  as well as is reasonably steady to database upgrades [10].

Algorithm 1 Feature Selection

Input: Graph database D, Discriminative ratio,

Size-increasing support function,

Maximum fragment size max L.

Output: Feature set F.

1: let F = { fФ }, DfФ = D, and l = 0;

2: while l <= maxL do

3: for each fragment x, whose size is l do

4: if x is frequent and discriminative then

5: F = F u {x} 

6: l = l + 1;

7: return F; 

 Algorithm 2 Search

Input: Graph database D,

Feature set F, Query q,

Maximum fragment size max L.

Output: Candidate answer set Cq.

 1: let Cq = D;

2: for each fragment x is subset of  q

and len(x) <=  maxL do

 3: if x € F then

4: Cq = Cq П Dx and return Cq.

 Algorithm 3 Insert/Delete 

 Input: Graph database D, Feature set F,

Inserted (Deleted) graph g and its id gid,

Maximum fragment size maxL.

1: for each fragment x is subset of  g

and len(x) <=  maxL do

2: if x € F then

 3: Insert Insert gid into the id list of x;

4:Delete;delete gid from the id list of x;

5.Return;

Rmat Algorithm

Inside this specific recursive system for the graph, mining discovering the attributes of genuine graphs which appear to continue more than several procedures [11]. We identify such “laws” as well as, more significantly, suggest a straightforward, parsimonious method, the recursive matrix (R-MAT) system, which could rapidly produce accurate graphs, recording the importance of every single graph in a mere a couple of variables. R-MAT immediately creates graphs using the neighborhoods inside of networks property. R-MAT can conveniently come up with convincing weighted, directed and bipartite graphs[13]. 

Proposed Algorithm

The  suggested  algorithm  is actually improved  in  overall performance  than  earlier algorithms such as for example  gIndex , Part Miner, gSpan & RMAT when it comes to of  grouping  and  looking around  including  DFSS  with  both  left  and  right connection, graph property with individual dependent query and connection property [12].

That contains the preceding procedures.

1. Development of nodes, feature of nodes, and connection between individuals nodes

2.  An assortment of property to be explored and arranging together with the assistance of relationship.

 3. Traversing towards a specific node which often requires being explored in simultaneously left  as  well  as  right  way  and  save  the  relationship  whenever  the pattern took place. 

Algorithm For Traversing  

Setup

Step 1 Create Graph Database

Step 2 Create Node

Step 3 Set Property of nodes

Step 4 Create Relationship

Step 5 Select p                 

 /* Property to be searched */

Step 6 Sort the graph by their relationship

Step 7 for Node position traverse <-  depth

Step 8    if  p == node.property

// if required property match

Step 9   S <- node-relationship 

// store relationship of     first match

Step 10   if node.left.relationship ==S

Step 11   display properties

Step 12   continue traverse down

Step 13   else

Step 14   if node.right.relationship ==S

Step 15   Display properties

Step 16   continue traverse down

Step 17   else

Step 18      traverse <- down next node

Step 19   end

Step 20   end

Step 21   if p==node.property

Step 22    repeat

Step 10 through18

Step 23   end

Step 24   end

Step 25   end

Comparitive Study & Discussion 

The proposed algorithm  after reviewed with previously mentioned existing algorithm works amazingly well ,  it keeps the data in a classified way, seeking happens in both directions left and also right,  graph   property taking into account client based query furthermore checks the connection whether it is coordinated, one to numerous or many to many relationships. The comparability of existing algorithm with proposed algorithm has appeared in Table 1.

A few snapshots have been taken during the graph database in order to demonstrate the graph database result. At figure 1, 2, 3.

Table 1: Comparison of existing algorithm with proposed algorithm

Feature	Part Miner	gSpan	gIndex	R-MAT	Proposed Algorithm
Sorting	Yes	Yes	No	No	Yes
Approach	Up-down	Up-down	Up-down	Up-down	Up-down
Search	DFSS	DFSS	DFSS	DFSS	Both ways Left to right, DFSS
Partition	Yes	No	Yes	Yes	No
Big  DB		Good			Good
Graph Property	No	No	Feature based	No	Yes
Relation	No	No	No	No	Yes

 

it  is  displaying  the  result  concerning  the  program code alongside  the  assorted  types  of  offenses  as well as  the  individuals  who  are engaging in violations plus beneath it information a few choices is also provided  through  selecting  any  of  these individuals  we  can easily  obtain  the  connection  of  that Fraud  that  might  assist  in  the  upcoming  calculation. Summary of graph database production is presented in figure 4.It is demonstrating the name and the information of different types of bank fraud we categorized.

Figure 1: Overview when database created Click here to View figure

 

 

Figure 2: Overview when database created  Click here to View figure

 

Figure 3: Overview when database created  Click here to View figure

 

Figure 4: It  is  showing  the  information  in  the  form  of  the  graph  nodes relationship with bank frauds classified in this system.   Click here to View figure

 

Figure 5: Displaying the ATM Frauds  nodes of  classified frauds  in this system.   Click here to View figure

 

Figure 6: Demonstrating single ATM Fraud case node of classified frauds in this Database with Node relationship.   Click here to View figure

 

Figure 7: Displaying the net bank Frauds nodes classified frauds in this system.   Click here to View figure

 

Figure 8: Displaying different view of graph of net bank Frauds classified frauds in this system.   Click here to View figure

 

Conclusion & Future Scope

In spite of the fact that the present  algorithm as of now performs entirely well, it can be implemented  on a regular basis frameworks to follow the pattern of Fraud ascent and fall in the offer financial sector and we can look at the present graph  of financial changes with the pattern present in graph database, so that on the off chance that it finds any similarity in the pattern it can force a security check over that specific transaction and foresee the future progress .This could be helpful in arranging the avoidance of a few wrongdoings which can add to the general population who gets influenced because of fraud. graph mining is a right now exceptionally dynamic research industry. The application zones of graph mining are across the board expanding from science and technology to web applications.

Acknowledgment

I am immensely indebted and owe my due regard to Dr. H. L. Mandoria, Professor, Information technology Department, Mr. Binay Kumar Pandey, Assistant Professor, Information Technology Department, Mr. Ashok Kumar & Mr. Rajesh Shyam Singh Assistant Professor, Information Technology Department & the members of my advisory committee for their persistent encouragement and support.

References

1. R. Agrawal and R. Srikant. Fast algorithms for mining association rules. In VLDB’94, pages 487–499, Sept. 1994.
2. Usama Fayyad, Gregory Piatetsky-Shapiro,   and Padhraic Smyth,”  From  Data  Mining  to     Knowledge  Discover Databases”, AI Magazine Volume 17 Number 3 (1996) (© AAAI)
3. D. J. Cook and L. B. Holder (2000) Graph-Based Data Mining, IEEE Intelligent Systems, 15(2).
4. A. Inokuchi, T. Washio, and H. Motoda. An apriori-based algorithm for mining frequent substructures from graph data. In PKDD’00, pages 13–23, 2000.
   CrossRef
5. X. Yan and J. Han. gspan: Graph-based substructure pattern mining. Technical Report UIUCDCS-R-2002-2296, Depart- ment of Computer Science, University of Illinois at Urbana- Champaign, 2002.
6. . Huan, W. Wang, J. Prins, Efficient mining of frequent sub graph in the presence of isomorphism,  in:  Proceedings  of  the  2003  IEEE International Conference on Data Mining (ICDM’03), 2003, pp. 549– 552.
7. Yan, X., Yu,  P.  S., & Han, J.  (2004). Graph indexing:  a frequent structure-based approach. In ACM SIGMOD international conference on management of data (SIGMOD’04) ACM, New York,  (pp.  335– 346).
   CrossRef
8. Hsinchun  Chen,  WingyanChung,  Jennifer  Jie  Xu,  Gang Wang  Yi  Qin  and  Michael  Chau,”  Crime  Data  Mining:  A General  Framework  and  Some  Examples”,  0018- 9162/04/$20.00 © 2004 IEEE
9. R.  Niewiadomski, J.  Amaral, and R.  Holte.A parallel external- memory frontier breadth-first traversal algorithm for clusters of work- stations. In IEEE ICPP, 2006.
10. JunmeiWang,  WynneHsu  Mong  and  Li  Lee  Chang  Sheng,”  “A Partition-Based Approach to Graph Mining”, Proceedings of the 22nd International  Conference  on  Data  Engineering  (ICDE’06)8-7695-2570-9/06 $20.00 © 2006 IEEE
11. Frank Eichinger, KlemensB¨ohm and Matthias Huber,” Improved Software Fault Detection with Graph Mining”, Appearing in the 6th International  Workshop  on  Mining  and  Learning  with  Graphs, Helsinki, Finland, 2008.
12. Sytske Besemer,” The impact of timing and frequency of parental criminal  behavior   and  risk  factors  on  offspring  offending”,  a Institute  of  Criminology  University  of  Cambridge,  Cambridge,  UK Version of record first published: 05 Nov 2012.
13. Justin J. Miller,”Graph Database Applications and Concept with Neo4j”,Proceedings of the  Southern Association for Information  System  Conference,  Atlanta,  GA,  USA March 23rd-  24th, 2013
14. Navneet Kumar Kashyap, Binay Kumar Pandey, H. L. Mandoria & Ashok Kumar,“A Comprehensive Study Of Various  Kinds Of Frauds & It’s Impact”, International Journal of Computer Science Engineering and Information Technology Research (IJCSEITR) ISSN(P): 2249-6831; ISSN(E): 2249-7943  Vol. 6, Issue 3, Jun 2016, 47-58

---

This work is licensed under a Creative Commons Attribution 4.0 International License.