In our last article, we talked abstractly about the danger of the human-based knowledge graph. The idea that knowledge workers, in their endless struggle to make sense of their organization’s data at the pace demanded of them, find that there’s no common ground to gather, collaborate on, or share their knowledge. In response, they end up siloing their insights onto pieces of paper scattered around their cubicle, into random files on their computer, or worse, into various corners of their own head.
We also positioned knowledge graphs as a proven solution to this problem—a way for knowledge workers of all technical abilities to explore their organization’s data, ask tough questions thanks to the unification of (un)structured data and AI/ML workflows, and bring their knowledge into the public square.
We’re going to get more specific now, using a GraphGrid customer as our example.
This financial consulting firm serves multinational organizations in analyzing and managing risk around financial sanctions from the U.S. Treasury, which change a few times every month. If the consultant’s clients fail to predict the sanctions change or implement a workaround fast enough, they’re subject to enormous fines from the U.S. government.
But even understanding the list of current sanctions means downloading and sifting through 21 different files and thousands of names. Accurately predicting which partners or customers the Treasury will add next means sifting through massive amounts of unstructured and unrelated data, whether that’s videos or social media, to uncover valuable knowledge.
Without clear answers on today’s and tomorrow’s sanctions from the U.S. Treasury, these multinational organizations can’t act fast enough to avoid massive federal fines. Enter the consultancy, which is responsible for giving these firms the knowledge they need to avoid trouble and streamline an already-complex network of financial entanglements.
Why a lack of connected, collaborative knowledge is so troubling
One obvious solution to this problem is for the consultancy to build an open-source catalog of people, places, and things the Treasury has already sanctioned and track their activities. The consultancy could convert this catalog of people, places, and things—plus the relationships between them—into valuable reports that help their clients sanctions before they happen.
For example, the consultancy notices that sanctioned Person A starts a new relationship with unsanctioned Business B. They flag this relationship in reports sent to the client, which can immediately investigate the risks and opportunities in replacing business B with alternative third parties. Now that they have new options available to them, they’re ready to take action if and when a new sanction is applied, preventing supply chain interruptions and avoiding damaging fines from the U.S. Treasury.
Building the database to support these reports will always be one of scale for this consultancy. They knew they could ask 100+ analysts to spend the first few hours scouring open-source data, from social media posts to information buried in the U.S. Treasury website, to build out their database. Their analysts are highly skilled at discovering, labeling, creating, and updating entities, but leadership quickly understood they couldn’t afford the consistent manual effort.
The limiting factor isn’t how much information they can gather, but instead handling the sheer volume of potential relationships between the people, places, and things analysts discover in their open-source research without assigning each analyst to a specific area of coverage.
And how to make those relationships productive for their customers.
Here’s an example of a nuanced relationship that an analyst might need to uncover and recognize in their work: Person X, who works for company Y, promoting a fundraising opportunity for non-profit organization Z, which has a history of dealings with sanctioned organizations. These connections might seem vague at best, but they generate knowledge that’s months ahead of new compliance standards—precisely what the consultancy’s customers pay handsomely for.
From the get-go, the consultancy understood the risk of human-based knowledge graphs. Analysts tend toward information they’re already familiar with, ignoring the larger picture, and siloing their insights from others.
Without a collaborative foundation on which they could share information and relationships, two of the consultancy’s analysts could be tracking opposite sides of this delicate chain of relationships, completely unaware that the other has an entire dossier of potentially relevant information.
What do you do with 400 new sanctions?
After recognizing their problem, this consultancy looked to GraphGrid CDP to automate the once-manual task of discovering, importing, and labeling all their open-source research inputs, like social media and news feeds. This dramatically simplified how the analysts identify relationships between the people, places, and things that matter to their clients, like new interactions between sanctioned people/organizations and third parties they’re in business with, such as lenders or supply chain partners. And by un-siloing knowledge from their many analysts and putting it into a single collaborative knowledge graph, the consultancy suddenly had new ways to capture and expose context, which is their most valuable commodity.
Read the full implementation story in our case study, From Weeks to Hours: How Knowledge Graphs and AI Transformed the Risk Management Business.
When the U.S. Treasury released 400 new sanctions into an already complex network of open-source information, the consultancy had the perfect opportunity to test their GraphGrid CDP knowledge graph.
Previously, their first step would have been to manually input the names of these 400 people and organizations, label their relationships, and connect them to existing knowledge using open-source research.
Instead, using GraphGrid CDP, the consultancy’s analysts set up automatic ingest of sanctions data when released by the U.S. Treasury. This process identified new people or entities and strengthened relationships using contextual clues, which meant analysts never had to capture and label data. Instead, they could focus on analyzing reports of only the most relevant results, with their knowledge graph surfacing the types of nuanced relationships their analysts might have missed during time-consuming manual operations of the past.
But most importantly, the analysts were freed from their silos and comfort zones thanks to GraphGrid’s single, collaborative knowledge graph. They could visually explore relationships—the edges, labels, and properties of a knowledge graph—and then query the 90% of data they hadn’t paid attention to before. Analysts with deep experience could quickly coordinate across their specific domains to provide an accurate and in-depth risk assessment report in a fraction of the time.
In just four hours, their team turned over the report to one of their clients, a multinational bank that must strictly adhere to the U.S. Treasury’s sanction list. The bank later told them their internal team would have needed several weeks to understand the inputs, relationships, context, and takeaways from the new block of sanctions.
Turn weeks into hours with GraphGrid CDP
Whether you’re doing open-source intelligence research, trying to understand and capitalize on changes to your supply chains, or detecting fraud in real-time, a connected and collaborative knowledge base is your next best tool for reducing complexity and creating a faster “time-to-knowledge” cycle for your entire organization.
If you haven’t already, you can follow in this consultancy’s footsteps by downloading GraphGrid CDP entirely for free, and with the full feature set enabled.
It’s everything you need to ingest data, analyze unstructured data and build relationships using natural language processing (NLP) and enable all your people, from data analysts to non-technical business users, to drive the kind of collaborative knowledge-building and -sharing that their job requires—and your organization demands.