Tree-Based User Agent Lookup for Mobile Device Identification

The patent segments the large look-up table into multiple smaller trees organized by device categories (e.g., smartphones, tablets, feature phones). Each tree contains a subset of user agent strings specific to that category. This segmentation reduces the search space from examining all entries in a large table to searching only relevant subsets, thereby maintaining identification accuracy while significantly reducing processing time.

The patent introduces a categorical dimension to organize user agent strings by device type (smartphones, tablets, feature phones, etc.). Instead of a flat, one-dimensional look-up table, the system creates a multi-dimensional structure where strings are grouped into trees based on device categories. This dimensional organization allows the system to handle diverse device variations without proportionally increasing complexity, as new devices can be accommodated by adding them to appropriate existing categories.

The patent implements partial matching within each tree structure, allowing the system to identify devices even when the user agent string does not exactly match any entry. The matching process examines portions of the string (prefixes, key segments) rather than requiring complete exact matches. This partial matching approach maintains high precision for known devices while significantly improving reliability by successfully identifying devices with string variations, typos, or minor modifications.