Heuristic Rule Generation for Spam Email Identification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current spam filtering technologies face challenges in efficiently identifying spam emails without compromising email content analysis, often resulting in false alarms and misclassification, especially when dealing with bulk spam mailing and the increasing sophistication of spam senders.

Innovation Solution

A system and method for generating heuristic rules that analyze email headers using statistical data and clustering algorithms to identify spam emails, employing MD5 hashing and regular expressions to create rules that distinguish between legitimate and spam emails without opening or accessing the email content, thereby maintaining confidentiality and reducing false alarms.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If content analysis is performed to identify spam emails, then detection accuracy is improved, but processing time increases significantly

Engineering Contradiction:
Improvespam detection accuracyVSAvoidemail processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the email into multiple components (headers, body, attachments, signatures) and analyzes each segment separately using different techniques. Header analysis uses statistical methods for quick evaluation, while body analysis is performed only on emails that pass the header filter, reducing overall processing time while maintaining detection accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary analysis of email headers before analyzing the email body. The header analysis includes statistical evaluation of sender information, routing paths, and message characteristics to quickly identify obvious spam. This preliminary action filters out many spam emails before the more time-consuming body analysis is required.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If email content is opened for analysis, then spam detection capability is improved, but user confidentiality and consent requirements are compromised

Engineering Contradiction:
Improvespam detection capabilityVSAvoidemail content confidentiality
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent extracts and analyzes specific features and characteristics from the email content (such as header fields, routing information, and statistical patterns) without requiring the user to open or fully access the email body. This extraction approach maintains confidentiality while providing sufficient data for spam detection.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces an intermediary analysis layer that processes email metadata and statistical features before any potential content access. This intermediary layer evaluates routing paths, sender patterns, and message characteristics to determine spam probability without directly exposing or requiring access to confidential email content.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If threshold is lowered to reduce false alarms, then legitimate email delivery is improved, but spam detection effectiveness deteriorates

Engineering Contradiction:
Improvelegitimate email delivery rateVSAvoidspam detection effectiveness
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent uses multiple evaluation parameters and thresholds instead of a single threshold. Different parameters (header statistics, routing characteristics, body content features) are weighted and combined to produce an overall spam probability score. This multi-parameter approach allows for more nuanced decision-making that reduces false alarms while maintaining spam detection effectiveness.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3716540B1System and method for generating heuristic rules for identifying spam emails
Publication Date: 2024.03.13 AO KASPERSKY LAB
  • EP3716540B1 patent drawingFigure 1
  • EP3716540B1 patent drawingFigure 2
  • EP3716540B1 patent drawingFigure 3

AI summary

Disclosed herein are systems and methods for generating heuristic rules for identifying spam emails. In one example, a method comprises, collecting and analyzing statistical data on contents of a emails to identify different types of content, including headers or hyperlinks, grouping the emails into clusters based on identified types of content, at least one cluster including groups of fields in the headers of said emails, selecting at least one most frequent combination of groups of data in each cluster, generating a hash from the at least one most frequent combination of groups, formulating at least one regular expression based on an analysis of hyperlinks corresponding to the generated hashes, and generating at least one heuristic rule for identifying emails containing spam by combining at least one hash and the corresponding regular expression, wherein at least one hash is from sequences of fields in the headers of said emails.