Mobile Device Attribute Detection via HTTP Headers and Social Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining mobile device attributes rely heavily on user-agent string databases, which can be resource-intensive and prone to inaccuracies due to the vast number of devices and the need for continuous updates.
Innovation Solution
A device-detection system that uses information from HTTP headers and user data, including social networking information, to determine mobile device attributes through models or algorithms, allowing for customized webpage delivery without relying on extensive databases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If user-agent string databases are used to determine mobile device attributes, then device identification can be performed, but the system becomes resource-intensive and prone to inaccuracies
Solution Approach 1:
The patent replaces the mechanical database lookup system with a machine learning-based detection system. Instead of comparing user-agent strings against extensive databases, the system uses trained models that analyze HTTP headers and social networking information to predict device attributes directly, eliminating the need for manual database maintenance while improving accuracy
Solution Approach 2:
The patent introduces social networking information as an intermediary data source between the HTTP request and device attribute determination. This additional information layer helps disambiguate device characteristics that cannot be reliably determined from user-agent strings alone, improving identification accuracy without requiring larger databases
2Adaptability or versatility
If extensive databases are maintained to cover the vast number of mobile devices, then device coverage improves, but resource consumption and update requirements increase
Solution Approach 1:
The patent replaces the resource-intensive database storage and lookup system with a compact machine learning model that has been pre-trained on diverse device data. The model can generalize to new device types without requiring database updates, reducing memory requirements and processing overhead while maintaining broad device coverage
Solution Approach 2:
The patent changes the approach from storing extensive device attribute data in databases to storing compact model parameters that can predict attributes. This transformation reduces the system's resource footprint while maintaining the ability to identify attributes across a wide range of devices through probabilistic prediction
3Productivity
If traditional database comparison methods are used, then device identification can be performed, but the system is slow and inefficient
Solution Approach 1:
The patent performs preliminary training of machine learning models offline using extensive device data. This pre-computation allows the system to make rapid predictions during actual device detection without performing time-consuming database comparisons, significantly reducing webpage customization time while maintaining high accuracy
Data Source
AI summary
In one embodiment, a method includes receiving a request for a webpage from a mobile-client system of a user, where the request includes an http-header, accessing information describing the user, determining the attributes of the mobile-client system based on the http-header and the information describing the user, and transmitting the webpage to the mobile-client system in response to the request, where the webpage has been customized based on the determined attributes of the mobile-client system.