AOH uses numerous traditional techniques for spam detection in combination with a variety of next-generation approaches for message analysis. Within this wide spectrum of tests, generally no single element will by itself classify a message as spam – thus avoiding the false positives associated with simpler approaches – while the breadth of analysis results in an industry-leading detection rate for junk email.
AOH’s technology for analyzing messages includes elements such as:
- Authenticity Checks including detailed header analysis, SMTP conversation details, message encoding and formatting, and other characteristics
- Message Fingerprinting to compare email signatures to frequently-updated public and internal databases of known spam messages
- An extensive, continually updated Heuristic Rule Set that encompasses message headers, body text, and other characteristics of both English-language and non-English messages
- Public and Private Blacklists of mail servers, relays, or networks known to be used by spammers
- Extensive URI Databases of unique elements such as URLs, phone numbers, and physical addresses known to be used by spammers
- Incorporation of Domain History and Reputation to increase the accuracy of blacklists and URI databases
- A self-learning Bayesian Engine that analyzes patterns of phrases in messages, and assigns mathematical probabilities for the presence of those phrases in junk mail versus legitimate mail
- Real-Time Message Source Analysis to assess whether an increased volume of mail flow is simply a legitimate high-volume mailing, or the result of a spammer hijacking or the use of a ‘zombie’ network
- Dynamic Feedback-Based Rules Optimization to leverage feedback from thousands of users as well as from many monitored legitimate and ‘spam trap’ email addresses
- User-Based Message Profiling that allows for refinement of message scoring for each user based on a history of his or her message traffic
- Customizable whitelists and blacklists which can be applied on an account-wide, organization-wide, domain-wide, or user-specific basis
Working together, these elements accurately detect a very high percentage of spam, phishing emails, and other unwanted messages – while minimizing the chances of a legitimate message being detected as spam. |
