Probabilistic User Visibility Encoding for Low-Memory CTV Ads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Connected TV video ads (CTV) lack feedback information for user awareness and engagement, leading to inefficient advertising strategies due to high memory requirements for storing user interaction data.
Innovation Solution
A method and system for encoding user visibility count with a low memory footprint using probabilistic algorithms and data structures, such as hashing functions, to track user device IDs and adjust visibility counts adaptively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If user visibility count is stored using traditional data structures, then measurement precision of user awareness is improved, but memory usage increases
Solution Approach 1:
The patent changes the parameter representation from exact integer counts to probabilistic floating-point estimates. Instead of storing precise visibility counts, the system uses probabilistic data structures that store parameters representing expected values and variances, enabling memory-efficient storage while maintaining sufficient measurement precision for advertising decisions.
Solution Approach 2:
The patent employs probabilistic data structures that can be easily created, modified, and discarded. These structures allow the system to maintain multiple user visibility counters with minimal memory overhead, treating the storage resources as disposable rather than permanent, which aligns with the transient nature of advertising campaign data.
2Measurement precision
If multiple counters per user are stored, then user engagement measurement is improved, but device complexity increases
Solution Approach 1:
The patent merges multiple user counters into a single probabilistic data structure per user. Instead of maintaining separate exact-count structures for each metric, the system combines them into unified probabilistic representations that capture multiple engagement dimensions while reducing the overall structural complexity and memory footprint.
Solution Approach 2:
The probabilistic data structures serve multiple functions simultaneously - they can represent visibility counts, engagement metrics, and various user interaction statistics within a single unified framework. This multi-functionality eliminates the need for separate specialized data structures for each measurement type.
3Manufacturing precision
If exact user visibility count is maintained, then advertising strategy accuracy is improved, but processing time increases
Solution Approach 1:
The patent applies partial action by maintaining probabilistic estimates rather than exact counts. The system accepts a controlled degree of imprecision in exchange for dramatically reduced processing time, using the principle that advertising decisions do not require absolute precision but rather sufficiently accurate estimates to guide bidding and targeting strategies.
Solution Approach 2:
The patent replaces the mechanical system of exact integer counting and storage with probabilistic computational models. Instead of incrementing and storing precise counts, the system uses probabilistic updates that leverage statistical properties to maintain accuracy estimates, substituting deterministic mechanical operations with probabilistic algorithms that are computationally more efficient.
Data Source
AI summary
The present disclosure provides a computer system (112). The computer system (112) performs a method for encoding user count with a low memory footprint. The method includes a first step of receiving real-time and adaptive frequency of user visibility. Further, the method includes another step of receiving a user device (106) id associated with one or more users (104). Furthermore, the method includes yet another step of encoding the user visibility count. The frequency of the user visibility is the number of times the computer system receives a request from a user device (106). The user device (106) id is a unique string of numbers and letters. The unique string of numbers and letters identifies the user device (106) associated with one or more user (104). The user visibility count is encoded by using one or more data structures and one or more algorithms.