Resource Classification via Access Measures

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for identifying malicious resources through blacklists and whitelists often result in high false negative and false positive rates, and are computationally expensive due to complex operations required for analyzing resource identifiers and resources themselves.

Innovation Solution

A resource classification system that assigns security classifications based on measurements of resource requests, generating resource access measures from request records and applying a classifier to determine the likelihood of a resource being malicious or benign, thereby reducing computational overhead.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If blacklist and whitelist approaches are used to identify malicious resources, then security classification capability is provided, but false negative and false positive rates are high

Engineering Contradiction:

Improveaccuracy of malicious resource identificationVSAvoidfalse positive and false negative rates

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The patent changes the parameters used for classification from static blacklist/whitelist membership to dynamic resource access measures including request frequency, number of unique clients, and temporal patterns. This parameter transformation enables more nuanced differentiation between malicious and benign resources, reducing false positives and false negatives by capturing behavioral characteristics rather than relying on binary listing approaches.

Inventive Principle:

Principle #35Parameter changes

2Reliability

If complex operations are performed for analyzing resource identifiers and resources themselves, then identification capability is provided, but computational cost is high

Engineering Contradiction:

Improveidentification capabilityVSAvoidcomputational overhead

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent extracts only the essential features needed for classification by focusing on resource access measures such as request counts and unique client metrics, rather than performing comprehensive analysis of resource identifiers and full resource content. This extraction approach maintains identification capability while significantly reducing computational overhead by processing only the most discriminative features.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent employs lightweight, easily computable metrics like request frequency and unique client counts that can be rapidly calculated and discarded after classification, replacing expensive long-running analysis operations. These simple measures provide sufficient discriminatory power for security classification while being computationally inexpensive and energy-efficient.

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

Data Source

PatentUS10122722B2Resource classification using resource requests

Publication Date: 2018.11.06 HEWLETT PACKARD ENTERPRISE DEV LP

AI summary

In one implementation, a resource classification system identifies a plurality of resource requests and generates a plurality of resource access measures based on the plurality of resource requests. Each resource request from the plurality of resource requests is associated with a resource from a plurality of resources by a resource identifier of that resource. Each resource access measure from the plurality of resource access measures is associated with a resource from the plurality of resources. The resource classification system applies a classifier to each resource access measure from the plurality of resource access measures to generate a classification result for the resource from the plurality of resources associated with that resource access measure, and assign a security classification to each resource from the plurality of resources based on the classification result for that resource.