Main functions of FinDet

1. Automatic and manual creation of rules to detect suspicious transactions. Detection of suspicious transactions: automatic and manual recording of transactions.
2. Reporting: creation and mailing reports to the Chief Inspector of Financial Information (GIIF), and follow-up reports.
3. Correction of rules.
4. Correction of transactions.

Automatic and manual creation of rules to detect suspicious transactions

findet2

Import of training data from the FinDet database. Data on

transactions are imported in a selective manner; data being essential for analyzing are collected, text data are converted into numerical ones.

Defining of classes of suspicious and "regular" transactions is made with application of GhostMiner data mining techniques. Here the clusterization module located in the Developer is being applied. Clusterization relies on joining of transactions into groups of similar properties. Each such a created group may constitute a starting point to determine a transaction class.

There are two methods of proceeding: 

• Creation of many classes of typical transactions based on transactions which are not doubtful (patterns of transactions); then each transaction which does not fit to any of typical classes would be automatically recognized as a suspicious transaction. 
• Creation of various classes of suspicious transactions; then in the future all transactions under analysis would be compared with classes of suspicious transactions; in case of compliance with one of them, they would be classified as suspicious transactions.

Detection of suspicious transactions

Detection of suspicious transactions is carried out in three ways: 

• by means of a manual marking of suspicious transactions by an operator (based on his knowledge)
• by means of rules of detecting of suspicious transactions defined by an authorized user (threshold value of transactions, definitions of related transactions)
• by means of automatically created rules of detection of suspicious transactions with application of data mining techniques

  

  finde3

Reporting 

• Reports including information on transactions submitted to the Chief Inspector of Financial Information (GIIF).
• Follow-up reports (GIIF has a right to inspect obliged institutions).

Management of rules 

• Edition / deletion / addition of rules in case when the rules were created by an authorized user.
• Import / export of rules in order to transfer rules between installations of the product in case of rules created by an authorized user and rules created automatically.

Correction of transactions 

• Upon importing a transaction from the banking system, supplementing the missing data that are required by GIIF without a possibility of changing data on transactions having been already imported from the obliged institution system.
• Adding transactions that are not found in the system.

Technical information

fidnet4

Safety  

• Multi-level system of operation.
• Analyst's level: possibility of defining users and creation of rules.
• Operator's level: possibility of suspicious transactions detecting.
• Manager's level: modification and verification of documents mailed to the Chief Inspector of Financial Information (GIIF).

Input data 

• Importing information on transactions from the obliged system.
• Manual transaction entering.
• Modification of data on transactions.