Consistent foundations
Multi-source records were prepared into consistent structures suitable for matching and relationship analysis.
Financial Crime & Risk Engineering
How OriginLabs Uncovered Hidden Financial Risk with Entity Resolution and Graph Analytics
Client: Major Canadian Bank (Anonymous)
OriginLabs helped connect fragmented counterparty and relationship data into a unified graph-based risk platform—giving investigators clearer context before the bank entered into a business relationship.
Review
Elevated review prioritySupporting reasons: indirect PEP connection · shared address · organizational relationship
Illustrative public visual. Labels and relationships are fictional; risk indicators support review and do not establish wrongdoing.
The risk blind spot
An isolated record could not reveal the network around a counterparty.
Before entering a new business relationship, the bank needed to understand not only who a party was, but how that person or organization connected to the wider network.
A counterparty could appear low-risk in one record while indirect relationships elsewhere introduced compliance or reputational considerations. The challenge was to make those connections visible without treating a signal as a conclusion.
Fragmented data
Relevant identity and relationship data lived across multiple systems.
Names, addresses, identifiers, ownership information, and related entities arrived in different formats. The same real-world entity could appear as several records, while important relationships remained split across disconnected views.
01Inconsistent identity records
02Duplicate people and organizations
03Indirect relationships hidden between systems
Entity resolution
From imperfect records to connected, traceable entities.
OriginLabs contributed to a graph-based risk intelligence platform that standardized source data, identified likely record matches, resolved those records into unified entities, and preserved the context investigators needed to examine the evidence.
Fuzzy identity matching
Likely matches could be identified even when names, addresses, identifiers, or company details were incomplete or inconsistent.
Unified entities
Related records were grouped into meaningful entity views while maintaining traceability to supporting evidence.
Nodes and edges
People and organizations became nodes; generic relationships such as directorships, ownership, and shared addresses became edges.
Public solution flow
From scattered records to connected risk intelligence.
This simplified sequence communicates the engineering approach without exposing source systems, internal architecture, matching thresholds, or proprietary scoring logic.
01Multiple Data Sources
02Data Standardization
03Fuzzy Matching
04Entity Resolution
05Graph Construction
06Risk Scoring and Reasoning
07Investigator Network View
Public technology set: Quantexa · Apache Spark · Scala · Python · SQL · AWS · Amazon S3 · AWS Glue · Amazon Redshift · OpenSearch
Explainable risk
A score accompanied by the reasons behind it.
Investigators needed more than a rank or label. The platform surfaced relevant connections and contributing factors so teams could understand what warranted attention and assess the evidence themselves.
Counterparty reviewElevated priority
- An indirect connection to a politically exposed person requires review.
- A shared address links records across separate entity views.
- An organizational relationship provides additional network context.
Illustrative reasons only. A score prioritizes investigation; it does not prove wrongdoing or replace investigator judgment.
Investigator network view
Direct and indirect relationships in one navigable context.
Representing entities as nodes and relationships as edges made multi-step paths easier to explore than isolated records or tabular searches. Investigators could follow the network, inspect contributing evidence, and decide what required closer review.
Navigate entity networks
Move from a central counterparty to connected people and organizations.
Trace indirect paths
See how risk context can emerge through several degrees of connection.
Review supporting reasons
Examine the relationships behind an assessment before making a decision.
Business impact
A more complete foundation for KYC and counterparty decisions.
Fragmented records unified into entity networks.
Indirect links made visible through graph analytics.
Scores supported by the relationships and factors behind them.
Are fragmented systems hiding critical risk?
OriginLabs helps teams turn disconnected identity data into connected, explainable intelligence that supports—not replaces—expert judgment.