TLDR

SMAT is releasing limited access to our Network Graph, a live point-and-click network graphing and link analysis tool for exploring cross-platform threats. By natively supporting queries across multiple platforms and allowing many connections, the tool turns flat data into the complex relations that better explain online threats. Our network graph is a cutting-edge tool for commercial, academic and independent researcher use. SMAT hopes this tool will empower users to spot malicious, multi-platform campaigns (eg. state-backed disinformation and astroturfing as well as grassroots harassment) and identify the key users and accounts driving them more easily.

Introduction

Throughout the last year, SMAT developers and contributors have been busy designing, engineering and building what we consider our flagship product. We began with the premise that online communities are not bound to one platform but instead distribute content across platforms in order to expand their audience and reach.

In order to fully capture the scope of any given online community’s activity, there would need to be a tool to represent all the cross-platform relationships between entities like “users”, “hashtags” and “links” for all the data sources SMAT actively crawls. All of that development has resulted in SMAT’s Network Graph, a live point-and-click network graphing and multi-platform link analysis tool, which is now available to the public under a limited release.

Functionality

To gather information on online communities, SMAT crawls the web by interacting with an API or parsing HTML and saves returned data (including individual posts, user profile objects, and channels) as flat files. These raw objects are stored in our back-end and are served via SMAT’s API tooling.

With Network Graph, SMAT is now able to show that crawled information in a graph structure, representing each data point while visually preserving its relationships with other data points. Still, to fully understand Network Graph’s potential impact and applications, it helps to relate it to how other online data relationships exist and are modeled by other applications and in other major, internet-driven industries.

When a user posts a hashtag, there is, in a sense, a relationship created between a user and the hashtag they posted. These kinds of data relationships exist all over the web, and representing them in graph form is already considered backbone functionality for many different web applications and technologies (ads and search, for example).

Until now, SMAT has lacked that capability—and as power users of our own analytic tools, our researchers found that pivoting from one data point to another while diving down investigative rabbit holes was time consuming and made it unnecessarily difficult to keep track of important information. 

The example above illustrates our new way of storing a Gab post using Network Graph. Usually, Gab posts contain information like who the author is, what channel it was posted in, and any other text-based information within the message such as a hashtag or a URL.

Previously, this data was both stored and represented as a single flat blob of JSON. Though we’re still storing data the same way, we can now visually represent the relationships between objects within the JSON blob as a networked graph to make browsing and analysis easier.

In Network Graph, we represent each JSON object as a node and each relationship between them as a line, each of which is defined and labeled along its edge, which is illustrated in the figure above. By showing the data this way, we can now visually intuit the following information and context:

• “Charlie The Cat” posted in the “All kinds of ANIMALS!!!!” group

• “Charlie The Cat” posted a link their Gab TV channel

• “Charlie The Cat” posted the hashtag “#cats”

• The hashtag “#cats” and URL to the Gab TV channel appeared in the same post

Additionally, Network Graph can relate and connect nodes across different sources and social media channels. In the example above, we traced the relationships between different online actors in a pro-Russia disinformation campaign by beginning with the infamous Telegram channel “WarFakes.”

The result was a very simple graph, from which we could easily draw some analytical insights:

• [A] We can see that the URL “warfakes[.]com” was posted in the Telegram channel “warfakes”

• [B] We can see that the URL “warfakes[.]com” was posted by several users in many places including the website LBRY, social media network Gab and subreddit “thedonald”

• [C] We see that one user in particular, “libman”, has a profile on additional channels and platforms including Parler and VK

Specs

To build this tool, we followed several technical specifications and design principles to drive value add. As a result, Network Graph is:

Interactive and Customizable

• Users can interact with the tool and build their own networks. There is no preset layout. 

Dynamic and Live-Updating

• Data will be different each time users interact with the tool because it is updated live. Every time a user runs a pivot, the information provided will be the latest that SMAT has crawled and stored.

Browser-based

• Users don’t need any third-party application, technical expertise, or special hardware to interact with or use Network Graph. All they need is a web browser!

Conclusion

In all, our hope is that Network Graph will make online research and analysis much more efficient, intuitive and rewarding, both internally at SMAT and externally for select independent scientists, researchers, and journalists. We will also be dropping guides, case studies and other insights we derive from our own in-house investigations using Network Graph, so stay tuned throughout the year for additional reporting and updates as we hit more of our milestones.

So far, our 2023 plans for Network Graph include: 

• Releasing an open-source version of the tool, including back-end and front-end code

• Deploying and maintaining a limited, public-access version of the tool 

• Building a tight community of researchers interested in beta testing the tool from the very beginning, including the current limited release

Finally, please get in touch at info@smat-app.com if you’d like to be considered for beta testing this release!