Exploring Russian Election Interference with a Knowledge Graph

We built a visual exploration tool and AI assistant for analyzing Russian interference in the run-up to the 2024 US elections.

Exploring Russian Election Interference with a Knowledge Graph
💡
Try the following:

1. Zoom into the graph using your mouse scroll wheel.
2. Hover over a node and read its summary.
3. Hover over an edge to read the fact derived from the relationship between adjacent nodes.
4. Click on a red Episodic node to read the article from which the related nodes and edges were extracted.

Graphiti is an open-source Knowledge Graph library at the core of Zep's memory layer for AI agents. Graphiti autonomously builds dynamic, temporally-aware knowledge graphs representing complex, evolving relationships between entities. In the run-up to next week's US election, we've used Graphiti to power our Russian Influence Operations Explorer: a knowledge graph visualization and related Q&A bot for exploring Russian state operations to influence US election outcomes.

💡
Graphiti helps you create and query Knowledge Graphs that evolve over time. A knowledge graph is a network of interconnected facts, such as “Kendra loves Adidas shoes.” Each fact is a “triplet” represented by two entities, or nodes (”Kendra”, “Adidas shoes”), and their relationship, or edge (”loves”).

We built the graph using a range of sources, including US DOJ indictments, US and foreign government research, non-governmental organization research, and media articles, to offer users a detailed view of these operations. Using our graph, you can explore which US organizations and individuals are implicated in these efforts and learn how private entities (such as OpenAI, Meta, and others) have responded to these challenges.

Two operations stand out in this dataset: Doppelganger and Tenet Media. The Doppelganger operation's multinational scope required us to draw from both US and European sources to capture its full impact. For Tenet Media, we used diverse sources, ranging from federal government reports to Variety magazine's industry coverage, to illuminate the company's collapse and its broader implications for the media landscape.

0:00
/0:37

How the Explorer works

Graphiti ingests data as a series of episodes, with a single data artifact constituting an episode. Nodes extracted from an episode's data will have an edge relationship with the Episodic node (marked in the graph in red). In our implementation of Graphiti, the red nodes include a reference to the source article or paper. Click on one to see where the graph entities came from!

Let's walk through an example research task. We want to explore the Russian state's use of sock puppet media and non-profit groups and ask the assistant how Russia uses these. Following a second or two of research, the assistant focuses the graph on relevant nodes and edges, and responds with a summary.

Exploring Russian Sock Puppet Media Sites

We notice an interesting organization, the Patriots Run Project, and zoom in.

Exploring the graph, we see that the fake social media accounts used in Patriots Run Project operation were acquired from entities or individuals in Bangladesh.

We asked the assistant for more information and learned that Russia was using the project to influence US conservatives.

Sources are important. The assistant will provide a list of sources used to answer your question. We currently provide the top 5 sources, but all sources for nodes or edges in the graph are available by clicking on a related red episodic node.

0:00
/0:06

💡
Want to learn how we built this app? Visit Building A Russian Election Interference Knowledge Graph.

Next Steps

  1. Visit the Explorer Application.
  2. Read more about Graphiti.

Bibliography

This project uses the sources listed below. They were selected based on their authority and diversity, and were available online as of Monday, October 28, 2024. (UPDATED 11/04/2024).

Government Agencies

Non-Governmental Entities

Media Organizations