IP2D Resolution System Using ML Source Vote Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current IP-to-Domain (IP2D) resolution systems often provide inaccurate mappings, leading to incorrect identification of organizational interests, as domain mapping sources may incorrectly associate multiple domain names with the same IP address, resulting in flawed analytics and targeting strategies for web content publishers.
Innovation Solution
An IP2D resolution system that generates unique features such as IP-level, domain-level, and source vote features to train an ML model predicting the most likely associated domain with an IP address, improving accuracy and efficiency in content consumption monitoring and analytics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional domain mapping sources are used to map domain names to IP addresses, then the mapping process is simple and fast, but the mapping accuracy deteriorates because multiple different domain names may be incorrectly mapped to the same IP address
Solution Approach 1:
The patent segments the domain mapping problem into multiple independent feature components: IP-level features, domain-level features, and source vote features. Each feature type captures different aspects of the mapping relationship, and their combinations enable more precise domain identification while maintaining systematic organization and manageable complexity
Solution Approach 2:
The patent transitions from traditional single-source domain mapping to a multi-dimensional approach by incorporating diverse feature types (IP-level, domain-level, source vote) and multiple domain mapping sources. This dimensional expansion enables more accurate disambiguation of domains sharing the same IP address
2Measurement precision
If multiple domain mapping sources are used to improve mapping accuracy, then the precision of domain identification improves, but the computational resources required increase
Solution Approach 1:
The patent performs preliminary action by pre-computing and storing IP-level features, domain-level features, and source vote features from multiple domain mapping sources. These pre-computed features are cached and can be quickly retrieved during domain identification, avoiding repeated heavy computations while maintaining high accuracy
Solution Approach 2:
The system implements self-service through automated feature extraction and model training processes. The machine learning model automatically learns from the pre-computed features and performs domain identification without requiring manual intervention, reducing operational computational overhead while maintaining high precision
3Measurement precision
If machine learning models are trained with multiple feature types to improve mapping accuracy, then the precision of IP-to-Domain resolution improves, but the device complexity increases
Solution Approach 1:
The patent segments the feature set into three distinct categories (IP-level features, domain-level features, source vote features), each with its own extraction and processing pipeline. This segmentation allows independent optimization of each feature type and simplifies the overall model training process while achieving high resolution accuracy
Solution Approach 2:
The machine learning model is designed with multi-functionality to handle multiple feature types uniformly. The model architecture can process IP-level features, domain-level features, and source vote features through a unified training framework, reducing the complexity that would arise from separate models for each feature type
Data Source
AI summary
An IP-to-Domain (IP2D) resolution system predicts which domain is most likely associated with an IP address. The resolution system generates unique source vote features (FSV) from (IP, domain, source) data. The FSV features are used to train a machine learning model that predicts which domain is most likely associated with an IP address. The domain predictions can then be used to more efficiently process events, more accurately calculate consumption scores, and more accurately detect associated company surges.