Suremessage powered by Proofpoint

Proofpoint spam detection

Unrivaled Spam Detection Accuracy

The Proofpoint Spam Detection™ Module provides the most powerful approach for detecting and eliminating spam. The key to its unrivalled accuracy is the patent-pending Proofpoint MLX machine learning technology, a system developed by scientists and engineers at Proofpoint's Anti-Spam Laboratory. Proofpoint has combined the most effective, traditional spam filtering methods with breakthrough machine learning technology, to deliver a system with the industry's highest effectiveness and lowest rate of false positives.

Multi-layered Spam Prevention for Maximum Effectiveness

Proofpoint Spam Detection uses a multi-tiered attribute extraction process that inspects more than 200,000 attributes of incoming email messages - including sender IP addresses, message envelope headers and structure - as well as unstructured content in the body of messages. These attribute-extraction layers include:

• Connection Level: The Proofpoint Email Firewall provides a stateful, first line of defense against spam by testing numerous connection-level data points including DNS, MX record verification and MLX Dynamic Reputation™ information. The Proofpoint Email Firewall also defends against directory harvest and denial of service attacks.
  
  
• Contextual Analysis: Examines the context of the message, using structural tests, English and foreign language inspection, pornography detection, URL inspection, targeted rules for detecting phish attacks and a corporate lexicon adapter to customize the solution to your industry.
  
  
• End User Configuration: Checks personal safe and blocked lists for valid and invalid senders.
  
  
• Administrator Customization: Checks global safe and blocked lists and any custom-created spam rules; global lists override end user lists.

All of the attributes detected within incoming emails are used by the MLX Anti-Spam Engine to ultimately assign a spam score which represents the probability that the message is spam. To stay ahead of evolving spam tactics, the engine is constantly and automatically kept up-to-date by the Proofpoint Dynamic Update Service.

Proofpoint MLX uses advanced machine learning techniques - such as logistic regression - to analyze more than 200,000 attributes in every email, providing extreme confidence in identifying spam messages.

Proofpoint MLX Provides Complete Confidence to Defeat Spammers

Proofpoint MLX technology goes far beyond the capabilities of competing anti-spam solutions. MLX is far superior to simple statistical techniques such as Bayesian filters - and it doesn't rely on signatures or fingerprinting techniques, which are easily fooled by spammers. It turbo charges traditional techniques with advanced machine learning technologies such as logistic regression and information gain analysis. The result is the highest spam detection rates in the industry. What does it mean to you? Complete confidence.

Classify Messages with High Confidence

The large number of attributes analyzed by the MLX Anti-Spam Engine ensures that messages are classified with a high degree of confidence. Most messages score very high or very low, with only 1.5% falling between 20 and 80 on a scale of 1 (an undeniably valid message) to 100 (assuredly spam). Confident scoring allows you to take decisive action - for example, automatically discarding spam messages before they impact your email servers, quarantining the absolute minimum amount of "probable spam" and delivering valid messages directly to end users.

The screenshot below illustrates the high confidence with which Proofpoint scores spam. Note the strong clustering of scores near 0 (definitely not spam) and 100 (definitely spam):

Competitors' products are often confused about how to classify messages - upwards of 40% of messages typically receive scores between 20 and 80, indicating uncertainty about the validity of the message. As a result, competing products often block valid messages as spam and many spam messages get through to inboxes. Misclassifications like these annoy end users and generate support calls, greatly increasing the total cost of solution.