Introduction

The Forta Network was established with a dual purpose: to empower developers and bolster security within the Web3 ecosystem. By monitoring real-time on-chain activity, Forta aims to detect threats, security related events and other noteworthy activities.

The escalating significance of Forta’s mission stands out; just a few years ago, security topics were barely on the radar at industry events. Today, they're front and center at major conferences, including the FCC and the recent Defy Security Summit. This shift underscores the critical role that security plays in building a reliable Web3 ecosystem - for Web3 adoption to truly take hold, both developers and users must have a sense of security.

The Forta Network is pioneering this security evolution with impressive metrics: 4,500 nodes running, 400 Forta bots actively monitoring the Web3 space, three premium data feeds, and over a 100 security experts building on top of Forta. Click here to apply for a one-month free trial of Forta’s threat intelligence.

“Utilizing Zettablock's platform significantly enhances the performance of Forta's machine learning infrastructure. Their GraphQL API makes the integration of large datasets with terabytes of historical data remarkably straightforward”, says Forta Researcher-in-Residence, Christian Seifert.

Real-Time On-Chain Monitoring

Forta employs a multi-layered approach to real-time scam detection, leveraging machine learning techniques (and other techniques) to stay ahead of threats. Central to the ML strategy lies a two-fold process: data collection and real-time inference. The ML models are first trained on static data to learn how to identify any potential victims. A few specialized bots also utilize graph neural networks to create similarity graphs around known scammer addresses in order to predict any future scammers, as shown below.

An example of Forta’s bots is the "scammer label propagation bot", which uses both graph neural networks and semi-supervised learning. Once an address is labeled as a scammer, the bot propagates this label to other addresses that have had interactions with the known scammer. This is done through label propagation techniques and high-quality data provided by ZettaBlock, allowing Forta to flag new addresses even before they engage in malicious activity.

[Source]

Leveraging ZettaBlock’s Historical Data and Real-Time API 

Training the Detection Bot with Historical Data

Forta's machine learning models require a wealth of historical data for training, and that's where ZettaBlock comes into play.

ZettaBlock's Data Lake provides a rich repository of all-time blockchain transactions, enabling Forta to train their models more effectively. By querying this data, Forta can create similarity graphs around known scammer addresses - which serve as the knowledge foundation for some of Forta's bots*, allowing them to learn the patterns and behaviors associated with scam activities.

*This is just for a subset of bots - not all of the bots need a lot of historical data and require ML models.

Real-Time Interface

First, in order to train the various models, historical data is required. This is where ZettaBlock's GraphQL API becomes invaluable. The APIs provide historical data on all addresses across a specific timespan. Forta then joins these addresses with their curated labeled data. After doing so, Forta uses this data, in order to train their various ML models. 

Once the models are trained, they are then used to do real-time inference with the latest on-chain data, that help provide quick and accurate assessments on potential scammers in the Web3 world.

To get started with your own Threat Detection Bot, visit our documentation page.

ZettaBlock VS an In-House Solution

Choosing ZettaBlock instead of an in-house solution was a strategic decision by Forta, driven by specific technical requirements. Forta required a unified backend capable of providing, high-quality, decoded transaction data. ZettaBlock's fully managed backend met these criteria, eliminating the need for Forta to build and maintain such complex infrastructure themselves.

Creating an in-house solution would have meant managing multiple data silos, each requiring its own pipeline for control and monitoring, as well as storing and maintaining over 10TB worth of Ethereum data (as well as other chains). The data size would also increase by many multiples, whenever Forta would want to expand their monitoring tools to new chains. This would have resulted in significant engineering challenges, not to mention the financial burden to store and process all of this data and pipelines. By leveraging ZettaBlock's robust backend, Forta can focus on their core competency - training and enhancing their machine learning bots for real-time scam detection - without getting bogged down by infrastructure complexities. This pivotal decision has hastened Forta’s market entry, all while cementing a safer and more dependable Web3 ecosystem.

Fun fact: Shield3 discovered ZettaBlock through Forta's GitHub repos. This ripple effect showcases how ZettaBlock not only empowers its direct clients but also influences the broader Web3 ecosystem. 

Conclusion

Together, Forta and ZettaBlock are on a mission to advance real-time security in the Web3 ecosystem. Forta's machine learning bots, trained on ZettaBlock's rich Data Lake, are revolutionizing scam detection. By opting into using ZettaBlock's fully managed backend over an in-house solution, Forta has streamlined its engineering operations, focusing on enhancing and improving the bots, rather than grappling with infrastructure challenges. Together, we are setting a new standard for blockchain security, making the Web3 world safer for all!

Want early access?!

Are you a developer working on similar Web3 security solutions? Don't reinvent the wheel! Let's collaborate and supercharge your project with ZettaBlock.

Sign up to ZettaBlock now - simply fill in this form!