knowfeat / inputs /SimECNY /README.md
youchengsong
Add README, domain knowledge inputs, and provenance cards
b34e287
|
Raw
History Blame Contribute Delete
11.3 kB

SimECNY

A simulator of wallet account, transaction, and closure for E-CNY.

Data Simulation

Wallet Creation Simulator

python open_wallet_distribution_based.py

Generated Files accounts.csv

Wallet Transaction Simulator

python generate_static_edges.py

python python simplified_main.py

Generated Files wallet_static_edges_distribution_based.csv wallet_temporal_transactions_1105_1450.csv

Wallet Closure Simulator

python generate_wallet_close_data.py

Generated Files new_close_3.csv

Data Description

I. Wallet Opening Module Attributes

The wallet opening data file records all relevant information when a wallet account is opened, including basic wallet information, identity authentication information, bank account information, region information, etc.

No. Attribute Name Chinese Description Data Type Description
1 wallet_id Wallet ID String Unique wallet identifier, numeric string
2 wallet_type Wallet Type Integer 1-Corporate wallet, 2-Personal wallet, 3-Merchant wallet
3 wallet_level Wallet Level Integer 1-Class I wallet, 2-Class II wallet, 3-Class III wallet, 4-Class IV wallet
4 wallet_open_cert Opening ID Card Number String 18-digit ID card number, can be empty (Class IV wallets do not have this field)
5 wallet_open_cert_expire ID Card Expiry Date String Format: YYYY-MM-DD, can be empty (Class IV wallets do not have this field)
6 wallet_open_date Wallet Opening Date String Format: YYYY-MM-DD
7 wallet_open_timestamp Wallet Opening Timestamp String Format: YYYY-MM-DD HH:MM:SS
8 wallet_open_tel Opening Phone Number String 11-digit phone number
9 realname_auth_time Real-name Authentication Time String Format: YYYY-MM-DD HH:MM:SS, can be empty (Class IV wallets do not have this field)
10 accociated_bank_account Associated Bank Account (Full) String Full bank account number, can be empty (Class III and IV wallets do not have this field)
11 bank_account_number Bank Account Number (First 4) String First 4 digits of bank account number, can be empty (Class III and IV wallets do not have this field)
12 init_balance Initial Balance Float Initial balance when wallet is opened, 2 decimal places
13 region_code Region Code String Region code used to identify the wallet's region
14 is_abnormal Is Abnormal Account Boolean True/False or 1/0, indicates whether it is an abnormal account
15 open_device Opening Device (MAC Address) String MAC address format (XX:XX:XX:XX:XX:XX), can be empty (if NaN, then opened at counter)
16 open_ip Opening IP Address String IP address, can be empty (if NaN, then opened at counter)

Notes:

  • Wallet Level Rules:
    • Class I and II wallets: Require all fields (bank card, ID card, authentication time, etc.)
    • Class III wallets: Do not require bank card binding, but require ID card and authentication time
    • Class IV wallets: Do not require bank card, ID card, expiry date, or authentication time
  • Counter Opening: Approximately 10% of wallets are opened at counters, in which case open_device and open_ip are empty
  • ID Card and Phone Number Reuse: According to configuration, some ID cards and phone numbers can be shared by multiple wallets

II. Transaction Module Attributes

The transaction data file records all transaction records between wallets, including basic transaction information, transaction patterns, risk indicators, device information, etc.

No. Attribute Name Chinese Description Data Type Description
1 tx_id Transaction ID String Unique transaction identifier
2 timestamp Transaction Timestamp String Format: YYYY-MM-DD HH:MM:SS
3 src Source Wallet ID String Unique identifier of the source wallet
4 dst Destination Wallet ID String Unique identifier of the destination wallet
5 amount Transaction Amount Float Transaction amount, 2 decimal places
6 transaction_motif Transaction Motif Type String Transaction type, such as: single_transaction, normal_small_high_freq,regular_large_low_freq, class4_laundering,merchant_laundering, online_laundering, small_amount_testing, etc.
7 motif_id Motif ID String Transactions sharing the same transaction pattern share the same motif_id
8 transaction_mode Transaction Mode String Transaction mode: single_transaction, forward, fan_out, fan_in, one_to_many, many_to_one, many_to_many
9 is_risk Is Risk Transaction String '0' indicates normal transaction, '1' indicates risk transaction
10 is_src_victim Is Source a Victim Integer 0 or 1, indicates whether the source is a victim account (for abnormal transaction analysis, mainly based on IP and device changes; if different from opening IP/device, it's fraud; if same, it's induced participation)
11 src_bank_account_number Source Bank Account Number String First 4 digits of source bank account number
12 dst_bank_account_number Destination Bank Account Number String First 4 digits of destination bank account number
13 src_wallet_level Source Wallet Level String Source wallet level (1, 2, 3, 4)
14 dst_wallet_level Destination Wallet Level String Destination wallet level (1, 2, 3, 4)
15 src_device Source Device (MAC Address) String Source MAC address, format: XX:XX:XX:XX:XX:XX
16 src_ip Source IP Address String Source IP address
17 dst_device Destination Device (MAC Address) String Destination MAC address, format: XX:XX:XX:XX:XX:XX
18 dst_ip Destination IP Address String Destination IP address
19 interval Time Interval Float Time interval (seconds) between adjacent transactions within the same motif_id
20 hour Transaction Hour Integer Hour when transaction occurred (0-23)

Notes:

  • Transaction Motif Types (transaction_motif):
    • Normal transactions: single_transaction, normal_small_high_freq, regular_large_low_freq
    • Abnormal transactions: class4_laundering, merchant_laundering, online_laundering, small_amount_testing (includes is_src_victim victim money laundering patterns)
  • Transaction Modes (transaction_mode):
    • single_transaction: Single transaction
    • forward: Forward transaction (appears in pairs)
    • fan_out: One-to-many transfer out
    • fan_in: Many-to-one transfer in
    • one_to_many: One-to-many (includes both transfer in and out)
    • many_to_one: Many-to-one (includes both transfer in and out)
    • many_to_many: Many-to-many (includes both transfer in and out)
  • motif_id: All transactions sharing the same transaction pattern share the same motif_id, used for correlation analysis
  • is_src_victim: Used to identify whether the source is a victim account in abnormal transactions (e.g., fraud scenarios)
  • Device and IP Information: If a wallet does not have device/IP information at opening, it will be randomly generated during transactions

III. Wallet Closure Module Attributes

The wallet closure data file records relevant information when a wallet is closed, including closure time, closure channel, risk indicators, etc.

No. Attribute Name Chinese Description Data Type Description
1 wallet_id Wallet ID String Unique wallet identifier, corresponds to wallet_id in the opening module
2 zs_id Closure ID String Unique closure record identifier, format: ZS000001, ZS000002, etc.
3 zs_channel Closure Channel String Closure channel, such as: mobile_app, web, counter, etc.
4 zs_timestamp Closure Timestamp String Format: YYYY-MM-DD HH:MM:SS, time of closure operation
5 wallet_open_tel Opening Phone Number String Phone number used when wallet was opened (for correlation analysis)
6 is_zs_laundering Is Abnormal Closure Integer 0 indicates normal closure, 1 indicates abnormal closure (money laundering related)

Notes:

  • Closure Time (zs_timestamp): Closure time should be later than the wallet's last transaction time
  • Closure Channel (zs_channel): Identifies the channel through which the user performed the closure operation
  • Abnormal Closure (is_zs_laundering):
    • Priority marking for wallets that meet both conditions: is_abnormal=1 and participated in abnormal transactions
    • Secondary marking for wallets that participated in abnormal transactions
    • Tertiary marking for wallets with is_abnormal=1
    • Finally, select other wallets based on risk scores
  • wallet_open_tel: Retains the phone number from opening time, used to analyze the historical account opening situation of phone numbers

IV. Data Relationships

  1. Opening Module and Transaction Module:

    • Linked through wallet_id
    • The src and dst fields in the transaction module correspond to wallet_id in the opening module
    • The src_wallet_level and dst_wallet_level in the transaction module correspond to wallet_level in the opening module
  2. Opening Module and Closure Module:

    • Linked through wallet_id
    • The wallet_id in the closure module corresponds to wallet_id in the opening module
    • The wallet_open_tel in the closure module corresponds to wallet_open_tel in the opening module
  3. Transaction Module and Closure Module:

    • Indirectly linked through wallet_id
    • Closure time should be later than the wallet's last transaction time in the transaction module