Problem

As retail, financial, and social activities move online, they generate new types of digital fingerprints such as email addresses, locations, device types, and operating system and browser versions. These digital fingerprints can provide rich intelligence to aid in detecting sophisticated attacks, but turning them into features or signals that can be used in rules engines or machine learning models requires extensive domain knowledge. Existing third-party services often provide signals on only one type of digital fingerprints at a time or provide simple black or white lists of them. Effectively leveraging digital fingerprints remains a challenge.

Feature Extraction Requires Deep Domain Expertise

Effectively leveraging digital fingerprint data (such as IP addresses, email addresses, OS versions, phone prefixes, or location) requires deep domain experience. Rules engines and machine learning models can not use these data types directly until they have been properly converted into features.

Ineffective Reputation-Based Systems

Existing solutions often provide coarse-grained reputation signals based on the history of a digital fingerprint (e.g., a device ID). These history-based reputations are often unreliable (e.g., due their dynamic natures such as dynamic IPs) or of limited coverage (e.g., unobserved before or new versions).

Unable to Analyze Patterns Across Multiple Data Sources

Existing solutions generally compute signals from each data type independently, ignoring the rich intelligence from the combinations of multiple digital fingerprints. For example, a user whose location is in the US but whose device is a phone that is sold only in China would not be marked suspicious by a digital fingerprint solution that looks at device type or locations only.

Solution

The DataVisor Global Intelligence Network leverages deep learning technologies to provide real-time, comprehensive digital intelligence based on the industry’s widest set of digital data, including IP addresses, geographic location, email domains, mobile device types, operating system, browser agents, phone prefixes, and more.

By analyzing the connections between these data points and not just each in isolation, DataVisor is able to provide fine-grained signals and reputations scores that can be either consumed directly in detection or to enhance the detection of rules engines and machine learning solutions.

Benefits

Extensive Domain Knowledge

DataVisor’s extensive domain experience in combating a variety of fraud, abuse, and money laundering activities allows us to create powerful features from a wide set of digital data.

Uncover Patterns Across Multiple Features

DataVisor’s Global Intelligence Network analyzes the connections between features, not just each feature in isolation. This allows it to determine when certain combinations are suspicious, improving accuracy and reducing false positives.

Analyze Each Feature in Detail

DataVisor’s Global Intelligence Network goes beyond history-based reputation scores, and outputs enriched signals such as populations size, frequency, diurnal patterns, the usage among specific types of customers, and more.

Real-Time Updates from DataVisor’s Network of 2B Users

DataVisor’s Global Intelligence Network is powered by aggregated signals from billions of user accounts worldwide, providing latest signals in real time as newest attacks patterns are discovered in our Network.

Architecture

The DataVisor Global Intelligence Network is one component of the DataVisor Detection Solution and works in concert with the Unsupervised Machine Learning Engine, the Supervised Machine Learning Module, and the Automated Rules Engine.

Ready to Enhance Your Detection with Unsupervised Machine Learning?

Request An Assessment