Machine Learning Relevance Scoring for Creative Content

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods fail to determine a multidimensional signature of creative content and calculate its relevance to a set of users without requiring user responses to the content or its modifications, and lack an interactive interface for real-time feedback on relevance changes.

Innovation Solution

A machine learning system that retrieves behavioral data, extracts features, builds models to predict content signatures, and compares user and content signatures to assess relevance, with an interactive interface for modifying content and providing real-time feedback on relevance changes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional user response measurement methods are used to determine creative content relevance, then measurement accuracy is improved, but time consumption and system complexity increase significantly

Engineering Contradiction:

Improverelevance measurement accuracyVSAvoidtime for sending and measuring user responses

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system pre-computes multidimensional signatures for users based on their historical behavioral data before actual creative content is presented. These signatures include user preferences, demographics, and engagement patterns that are calculated in advance and stored for rapid relevance assessment without requiring real-time user interaction.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary relevance scoring system that acts as a mediator between creative content and users. This system uses pre-computed user signatures and content analysis to generate relevance scores without directly engaging users, thereby eliminating the need for time-consuming A/B testing and user response collection while maintaining measurement accuracy.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If A/B testing with multiple content variants is used to improve creative content, then content effectiveness is improved, but the number of required user samples and time for iteration increase

Engineering Contradiction:

Improvecreative content effectivenessVSAvoidcontent iteration speed

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The system implements a feedback mechanism that uses pre-computed user signatures to immediately evaluate how different creative content variants would perform with specific user segments. This allows marketers to receive instant feedback on content effectiveness without waiting for actual user responses, enabling rapid iteration while maintaining reliable effectiveness measurements.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

User preferences and behavioral patterns are pre-analyzed and stored as signatures before content testing begins. This preliminary action allows the system to simulate content performance across different user segments instantly, eliminating the need for large-scale A/B testing while maintaining the reliability of effectiveness measurements.

Inventive Principle:

Principle #10Preliminary action

3Measurement precision

If detailed user behavioral data collection is implemented, then relevance prediction accuracy is improved, but data privacy concerns and system complexity increase

Engineering Contradiction:

Improverelevance prediction accuracyVSAvoidsystem complexity for data management

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system extracts only the essential features from detailed user behavioral data to create condensed multidimensional signatures. These signatures capture the most relevant user characteristics for content relevance prediction while discarding redundant information, thereby maintaining prediction accuracy while reducing data management complexity and privacy concerns.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies different levels of data processing to different user characteristics based on their relevance to content prediction. Sensitive personal information is processed at a higher level of abstraction while less sensitive behavioral patterns are retained with more detail, optimizing the balance between prediction accuracy and privacy protection while simplifying data management.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS11778049B1Machine learning to determine the relevance of creative content to a provided set of users and an interactive user interface for improving the relevance

Publication Date: 2023.10.03 PINPOINT PREDICTIVE INC

AI summary

One aspect of the invention is generating for a client (a marketer) a measure of relevance of online creative content for online marketing to a set of target users. Another is an interactive user interface for displaying such a measure. Another aspect of the invention is determining a signature of creative content and of components thereof. The measure of relevance is based on profiles of the targeted users and on the signature of the creative content. Another aspect of the invention providing a mechanism to improve the relevance by modifying one or more components of the creative content with respective alternate components, thus suggesting to the client modifications to the content together with a measure of how such modification changes the relevance. Another aspects is providing for an operator the user interface for viewing in real time the effect of such modifications.