Visual Crime Network Investigations Across Billions of Accounts and Messages
Explore how DataWalk’s Graph & AI platform combats sophisticated criminal activities like fraud and human exploitation across billions of accounts.
CASE STUDY
A leading global online platform, with a user base exceeding several billion accounts, faced a critical challenge: combating sophisticated criminal activities like selling illicit goods, human exploitation, fraud, and money laundering. Investigators were forced to contend with massive, complex data, making it difficult to uncover the true scope of these illicit networks.
Context and Results
The crimes they were facing were purposefully obfuscated, making them complex to investigate within such a vast dataset. Following a lead required traversing through many intermediaries, and the data volume grew rapidly with each step. Their existing solution couldn’t cope with this combination of depth and data size and left investigators with two broken workflows:
- Investigate widely but too superficially: Splitting a complex query into smaller, simpler ones to fit system limitations. This was slow and meant missing non-obvious connections deeply embedded in the data while trying to stitch results back together.
- Investigate deeply but too narrowly: Selecting a random data sample and designing custom code for analysis with IT. This was a tedious back and forth process where investigators had to hope that their query would execute successfully and take them down the right path, or else they had to start over with a new sample.
During a joint workshop, DataWalk’s scalable Graph & AI platform demonstrated an unprecedented capability to visually explore complex and deeply hidden connections while maintaining the full scope of the multi-billion entity data universe.
Unconstrained Data Exploration & Unified View: DataWalk enabled a unified view of all relevant data, allowing investigators to freely explore exponentially growing networks and complex relationships across the full multi-billion entity data universe without technical constraints.
Iterative Query Refinement & Composite AI: DataWalk demonstrated how investigators could refine their queries through iterative exploration, adapting their approach as new relationships and patterns emerged, powered by DataWalk’s composite AI which combines multiple advanced analytical techniques.
Scalability for Billions of Accounts: DataWalk’s capabilities included effortlessly traversing billions of accounts and their connections, allowing for the identification of sophisticated illicit networks and ensuring rapid adaptation to new challenges with unlimited computing power.
The Challenge: Stalled Investigations at an Unmanageable Scale
With a user base of several billion accounts and managing billions of daily messages, this platform was a constant target for illicit activity. Their dedicated team of hundreds of investigators frequently responded to law enforcement requests, but their methods were slow and limited. A single complex case could take weeks to resolve.
The core issue was not just the volume of data, but the inability to analyze relationships across multiple degrees of separation. While direct connections were easy to find, the real challenge was in tracing indirect relationships several hops away without hitting application data limits.
This meant investigators could not effectively explore the full scope of illicit networks. They were often stuck, unable to even formulate the right questions because the true extent of the activity was unknown.
Combating sophisticated criminal activities
Massive and complex data
Inability to analyze deep, indirect relationships
The Solution: Investigative Agility at Scale with DataWalk
DataWalk redefined how the customer analyzed vast, interconnected datasets. During a joint engagement, DataWalk rapidly built a unified knowledge graph from billions of records, enabling investigators to analyze data contextually and uncover hidden relationships using its powerful inference engine. A single suspicious entity could connect to seemingly unrelated networks through a chain of intermediaries, but identifying this required sophisticated graph traversal and pattern recognition to uncover hidden risk pathways, and this was beyond the scope of their existing tools.
A critical differentiator was DataWalk’s fuzzy matching engine, purpose-built for large-scale entity resolution. It accurately linked similar usernames and aliases—even with anagrams, symbols, or nickname variations—across billions of data points.
Instead of writing complex custom code, investigators could build modular, step-by-step queries without defining full paths in advance. This “brick-by-brick” method leveraged DataWalk’s optimized computation model to maintain low latency, even across massive graphs.
For example, using the entire >100-billion-element graph, investigators could instantly expand their view from a single suspicious account to reveal:
- 300 connected accounts
- 3 traded artifacts
- 900 associated accounts
- 60,000 IP addresses
- 10,000 related users
— all without hitting graph size or traversal limits. This end-to-end capability enabled the client to move beyond legacy constraints and expose entire criminal networks with unmatched speed and precision.
Download free ebook “How DataWalk AI is Transforming Investigative and Intelligence Analytics”
DataWalk Platform
Join the next generation of data-driven investigations
Discover how your team can turn complexity into clarity fast.
