Choose where to look.
Group transaction activity and apply your review criteria to identify accounts that need closer inspection.
Fraud investigation
Follow the connection.
Inspect the evidence.
Query transactions and trace related accounts, devices and beneficiaries. Give investigators a clearer view of the records behind a signal.
Evidence for review
Find shared connections. Give an investigator the records to review.
Illustrative applicationWhat you can build
Give the investigation more context.
Find the shared reference.
Follow declared relationships between accounts, devices and beneficiaries to explore connected activity.
Keep the evidence attached.
Retain record IDs, timestamps and query context so an analyst can examine the reason a connection was surfaced.
The application workflow
From a signal to a reviewable case.
A shared identifier is a starting point. Query the context and inspect its source before deciding what it means.
Suggested record structure
Facts an analyst can trace
-
account_id -
device_id -
transfer_id -
beneficiary_id -
timestamp
Your schema, built around your application.
-
Select a candidate
Use SQL to examine transaction counts and amounts within a chosen window. Your application defines the thresholds for review.
-
Expand the context
Follow declared graph relations with bounded depth and result limits. Keep the source record behind each link.
-
Assemble the case
Present the evidence, timing and query parameters to an analyst. Route review and escalation through your case-management workflow.
How it fits together
The data in one place.
The workflow in your hands.
Explore the architecture 
OriginChainDBStore and retrieve.
Retrieves transaction records, aggregates and declared connections. Helps identify neighbors and bounded paths worth investigating.
Coordinate and decide.
Defines review rules, assembles cases and routes decisions. Analysts or application policies determine whether further action is appropriate.
A shared device or beneficiary does not establish fraud. Check timing, identity and source quality. Graph capability must be enabled, and traversal needs to stay within its documented query limits.
Build for real use
Start with a workflow.
Check the whole result.
Could the connection have an ordinary explanation?
Review shared households, expected merchants and other legitimate relationships. A graph pattern describes recorded facts, not intent.
Is the investigation bounded?
Start with a narrow candidate set and explicit depth and result limits. Handle query-budget failures before expanding the search.
Can the case be reviewed later?
Save the identifiers, parameters, timestamps and supporting records that were used. Keep review outcomes in your application-owned case data.
Build with OriginChainDB