Sketching Generator for Digital Campaign Predictions

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional campaign-forecasting systems face issues with excessive storage consumption and inflexibility, unable to handle high-dimensional, nested complex set operations, and assume conditional independence among targeting criteria, leading to inaccurate results.

Innovation Solution

The content-campaign-prediction system employs a running-average-tuple-sketch algorithm to generate sketches for clearing-bid values and bid-success rates from historical auction data, allowing for multi-dimensional set operations and reducing storage consumption, thereby providing more accurate and flexible predictions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional campaign-forecasting systems use sophisticated models for robust targeting criteria, then prediction accuracy is improved, but storage consumption increases excessively

Engineering Contradiction:

Improveprediction accuracyVSAvoidstorage consumption

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential statistical characteristics (clearing-bid values and bid-success rates) from historical auction data using sketching algorithms, rather than storing complete historical data. This extraction approach maintains prediction accuracy while dramatically reducing storage requirements by keeping only compressed statistical summaries.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transforms historical auction data into compressed statistical parameters (sketches) that capture the essential distribution characteristics. By changing the data representation from raw historical records to condensed statistical summaries, the system achieves both accuracy and storage efficiency.

Inventive Principle:

Principle #35Parameter changes

2Device complexity

If conventional systems assume conditional independence among targeting criteria, then computational complexity is reduced, but prediction accuracy deteriorates

Engineering Contradiction:

Improvecomputational complexityVSAvoidprediction accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent segments the analysis into separate independent sketches for clearing-bid values and bid-success rates, while preserving the ability to analyze their joint distribution. This segmentation allows efficient computation while maintaining the capability to model dependencies between targeting criteria through the tuple sketch structure.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent moves from analyzing individual targeting criteria in isolation to analyzing multi-dimensional tuples that capture joint distributions. By adding dimensional information about co-occurrences of targeting criteria, the system accurately models dependencies without excessive computational complexity.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

3Adaptability or versatility

If conventional systems process high-dimensional targeting criteria, then targeting robustness is improved, but processing speed decreases

Engineering Contradiction:

Improvetargeting robustnessVSAvoidprocessing speed

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent performs preliminary processing of historical auction data to generate sketches before actual campaign forecasting. By pre-computing and storing compressed statistical summaries in advance, the system enables rapid real-time predictions without reprocessing raw historical data during forecasting operations.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent creates compressed sketch representations that serve as efficient copies of historical data distributions. These sketch copies retain the essential statistical properties needed for forecasting while enabling much faster processing compared to using complete historical datasets.

Inventive Principle:

Principle #26Copying

Data Source

PatentUS11348130B2Utilizing a sketching generator to adaptively generate content-campaign predictions for multi-dimensional or high-dimensional targeting criteria

Publication Date: 2022.05.31 ADOBE INC

AI summary

The present disclosure relates to systems, non-transitory computer-readable media, and methods to generate sketches for clearing-bid values and bid-success rates based on multi-dimensional targeting criteria for a digital-content campaign and dynamically determine predicted values for the digital-content campaign based on the sketches. To illustrate, the disclosed systems can use a running-average-tuple-sketch to generate tuple sketches of historical clearing-bid values and tuple sketches of historical bid-success-rates from historical auction data. Based on the tuple sketches, the disclosed systems can determine one or more of a predicted cost per quantity of impressions, a predicted number of impressions, or a predicted expenditure for the digital-content campaign—according to user-input targeting criteria and expenditure constraints.