Mechanism of risk determination

Sift holds more than 16,000 feature values based on data accumulated worldwide. Risk judgment is performed by combining the necessary feature values for the data actually sent, such as the characteristics of the access source terminal, the entered e-mail address, member information, and delivery address. (scoring from 0 to 100)

We also combine several types of machine learning models and perform calculations with the most suitable engine. The learning model is optimized according to the customer's environment as well as the globally shared feature values.

Examples of features

• Normally, the user will enter the e-mail address registered as member information and the delivery address.
• If a malicious user creates multiple accounts with similar email addresses and attempts to send them to the same shipping address, Sift will use the similarities to identify identity and make a risk determination.
• The relevance of the terminal information being accessed and the behavior from creating an account to placing an order are also used as feature quantities. Unlike rule-based systems, it automatically finds features according to incoming data and calculates the degree of risk for each feature.

Sift global network