According to a report published by the Financial Supervisory Service, the scale of insurance fraud in 2014 was estimated to be around 600 billion KRW.

In response, the Financial Supervisory Service announced plans to enhance the analytical capabilities of the Insurance Fraud Detection System (IFAS) 

by incorporating social network analysis functions to combat insurance fraud.

How can SNA be utilized for detecting insurance fraud?

Recently, SNA has been gaining attention for tracking collusion crimes such as insurance fraud and stock manipulation.

Collusion crimes, which involve two or more conspirators, often employ highly sophisticated methods and involve large and complex datasets.

As a result, traditional approaches to crime tracking have shown limitations, leading to increased interest in using SNA for collusion relationship tracking.

Collusion relationship tracking refers to the process of collecting traces of crimes recorded in digital data, such as communication and financial transactions, 

identifying hidden collusion relationships within them, and analyzing these relationships based on the Social Network Analysis (SNA) methodology to understand the essence of collusion crimes.

The investigative technique used to apprehend terrorists following the 9/11 attacks in 2001 was SNA.

Then, what are the limitations of traditional collusion crime tracking methods, 

and what advantages does collusion relationship tracking through SNA offer?

Traditional collusion crime tracking methods are limited in that they focus solely on analyzing individual suspect profiles, making it difficult to determine collusion involvement.

Second, without specialized methodologies and tools for collusion relationship analysis, relying instead on general office software or basic visualization programs, 

the tracking process takes longer and risks missing critical connections.

In contrast, collusion relationship tracking through SNA offers the following advantages.

First, it structures relationships between actors as network data, allowing for a comprehensive understanding of the entire collusion network.

Second, it identifies the connection structure between suspects, helping to determine who the central figures are.

Third, it enables the identification of key suspect groups.

Finally, by utilizing specialized analytical systems, it significantly reduces analysis time.

Now, let's examine how it is being applied in practice through case studies in various fields.

Case Studies on the Application of Collusion Relationship Tracking Solutions

Case 1. Detection of Stock Manipulation Through Collusion

1. Background and Issues

Stock manipulation is difficult to execute alone and typically involves collusion between two or more individuals. 

The methods used are highly sophisticated, making it challenging to distinguish from legitimate transactions. Additionally, the vast volume of trading data makes detection even more difficult.

2. Solution

To address this issue, first, a data model is established to integrate and analyze various data sources, along with relationship-based analysis and visualization functions to identify potential collusion.

Second, an algorithm is implemented to identify suspicious patterns, enabling the early detection of investigation targets within large-scale data.

3. Application Effect

By modeling complex data as relationship-based data, an integrated analysis environment is established, enabling intuitive analysis and visualization. 

As a result, the time required to create a collusion network of suspects, which typically takes 1 to 3 months, has been reduced to less than a week.

Case 2. Detection of Insurance Fraud Through Collusion

1. Background and Issues

Insurance fraud refers to the act of multiple parties involved in an insurance contract colluding to deliberately cause or disguise an incident in order to illicitly obtain insurance payouts. 

Such insurance fraud is increasing, and large-scale cases often involve complex structures with many interconnected individuals, 

making it difficult to fully grasp the nature of the crime and posing significant challenges.

2. Solution

The insurance fraud detection solution for collusion relationship analysis can be categorized into property insurance, primarily focused on traffic accident investigations, and life insurance, 

which deals with personal insurance matters. Representative cases for each category can be examined.

First, in traffic accident property insurance, collusion relationship analysis is conducted between the at-fault and victim vehicles. 

Since traffic accidents occur randomly, there should be no pre-existing connections between the occupants of the at-fault vehicle and the victim vehicle. 

Therefore, such collusion relationships are identified and extracted based on cycle patterns.

Second, collusion relationship analysis is conducted among hospitals, insurance agents, and insured individuals in disease-related personal insurance. 

Identifying collusion between insurance agents and hospitals is crucial in detecting fraud related to specific illnesses. 

Therefore, by measuring the strength of collusion relationships formed between hospitals and insurance agents through intermediaries, it is possible to identify suspicious groups.

3. Application Effect

By analytically utilizing collusion "relationships" in large-scale data, suspicious groups can be identified more efficiently, 

reducing the time required for insurance fraud detection and improving the accuracy of analysis. 

Especially when cycle pattern detection analysis is implemented, suspicious groups can be periodically extracted. 

Additionally, by minimizing the risk of data loss during the expansion process of extracted suspicious groups, individual insurance companies can increase the amount of detected insurance fraud.

We have examined case studies on the utilization of the collusion relationship tracking solution.

Cyram's collusion relationship tracking solution does not analyze "new" data.

It simply analyzes the data "differently" from traditional methods.

As such, the value and effectiveness of data analysis can vary significantly depending on the analytical method used and how the data is interpreted.

Thank you.