Accelerating Intelligence Analysis with Entity Resolved Knowledge Graphs

At the Senzing Global User Conference, Vince Bridgeman, SVP of National Security Services at Redhorse Corporation, and Michal Bachman, CEO of GraphAware, demonstrated how entity resolved knowledge graphs (ERKGs) can transform the speed and precision of intelligence workflows. This presentation highlights how connecting siloed data and surfacing hidden risks leads to faster, more confident decisions, especially in high-stakes investigative environments.

They walked step-by-step through a case study in which Redhorse and GraphAware partnered with Senzing to analyze fraud risk in pandemic-era Paycheck Protection Program (PPP) loans. With the combined power of Redhorse’s deployment expertise and GraphAware Hume – a graph analytics platform enhanced by Senzing® entity resolution – they showed how ERKGs uncover non-obvious connections, deliver risk-ranked investigative leads with full data lineage, and surface broader patterns that elevate the signals that matter most.

Bridgeman set the scene by describing the scale of the problem. During the pandemic, more than 11 million loans totaling $800 billion were issued by over 5,000 lenders – a massive distribution effort carried out at unprecedented speed. That urgency created blind spots, which fraudsters quickly learned to exploit. The challenge was to assess potential fraud, generate actionable leads, and provide a defensible, explainable analytics framework.

Bachman explained how GraphAware Hume, combined with Senzing entity resolution, creates a unified analytical environment. The platform turns disconnected data into an entity resolved knowledge graph (ERKG), enabling users to:

• Identify hidden relationships between people, companies, and addresses.
• Ask high-level questions – such as who really owns what or who is connected to who – without needing to know the structure of the data in advance.
• Surface fraud risk through analyst-defined, context-aware detection patterns aligned with investigative priorities.

As Bachman notes during the talk, “Our mission is to help anyone find truth in data. The world is more and more connected, so the tools we use need to embrace connected data and provide the opportunity for different kinds of analysis to traverse networks efficiently – and really understand who is who and how things are connected.”

Unlike traditional entity resolution systems that require entity resolution rules to be written and maintained, the Senzing SDK is principle-based and handles resolution out of the box, enabling teams to focus their effort on defining risk patterns and investigative priorities.

Bridgeman describes how the team built a working system using publicly available PPP loan data, business registries, address validation services, and watch list sources. They resolved entities with Senzing, loaded the results into a Neo4j graph database, and applied fraud detection logic, developed in collaboration with loan fraud analysts, within GraphAware Hume to drive investigations.

The system detects patterns such as loan duplication, watch list hits, and suspicious registration activity. It also scores risk, with results aggregated in a dashboard used to prioritize investigative leads. Because the ERKG preserves source lineage and context, analysts can drill down to the original records, making findings both explainable and actionable.

Bridgeman describes how this shift plays out in practice: “A knowledge graph combined with powerful tools like entity resolution is essentially a way to automate that understanding of relationships, accelerate the understanding of patterns, so you can spend more time understanding the principles and making better decisions with your data.”

With the combined toolset, the team uncovered real examples of fraud hiding in plain sight, including:

• Loan stacking: A business received 23 separate loans for one reported job.
• Watch list linkages: Shared addresses exposed ties to sanctioned individuals.
• Phantom companies: Loans went to businesses that didn’t legally exist.
• Statistical outliers: Salary claims far outside typical distributions.
• Foreign ownership flags: Simple graph queries using UBO data revealed overseas control.

During the Q&A, one audience member asked: Can’t we just do entity resolution with a graph database? Both speakers addressed it head-on. Bachman acknowledged it’s technically possible, but it’s costly, time-consuming, and rarely robust: “The question is, do you want to spend two years fine-tuning it, or do you want something that works out of the box?”

Bridgeman echoed the sentiment, recalling a failed project where a client insisted on using an internal entity resolution service that only worked on 25% of the data. “When we have these problems to solve,” he said, “we start with technology that already works so we can move fast, solve real problems, and trust the results.”

Entity resolved knowledge graphs change both the speed and depth of investigations, shifting analysts from reactive data sifting to proactive understanding of risk. With ERKGs, teams can move from data ingestion to actionable insight in hours instead of weeks.

00:05 Introduction to the Redhorse and GraphAware mission and partnership

01:02 The power of connected data and knowledge graphs for intelligence analysis

06:17 Real-world fraud detection: the PPP loan case study

11:31 How ERKGs surface risk patterns and support investigations

15:34 Example: high salary variance as a fraud indicator

17:43 How knowledge graphs accelerate decision advantage

21:10 Addressing the challenge of DIY entity resolution with graphs

24:15 Lessons learned for future crisis response and fraud prevention

Schedule a demo today and learn how entity resolved knowledge graphs are empowering intelligence analysts.

For more information, please check out:

8 Challenges for Enterprises Building Entity Resolution

Build or Buy Entity Resolution: Which Option is Best

Related Videos