ML Recommender System for Network Content Delivery Optimization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Content providers face challenges in optimizing their online content delivery campaigns as they often fail to reach the intended audience, making it difficult to determine the effective factors to change, leading to unpredictable results when adjusting campaign parameters.

Innovation Solution

The use of machine learning techniques, specifically decision trees and gradient boosting, to generate personalized recommendations for content delivery campaigns by predicting performance based on campaign attributes, allowing content providers to improve delivery, conversion rates, and audience reach.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If content providers manually adjust campaign parameters based on guesses, then they can make changes to campaign settings, but the results are unpredictable and delivery performance does not improve

Engineering Contradiction:

Improvecontent delivery performanceVSAvoidprediction accuracy of delivery metrics

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent replaces manual guessing and mechanical adjustment of campaign parameters with an automated machine learning system. The gradient boosting model automatically analyzes historical data and predicts delivery metrics, eliminating the need for human intuition-based adjustments and providing data-driven recommendations for optimization.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system implements feedback by continuously analyzing actual campaign performance data against predicted metrics. The machine learning model learns from historical outcomes and adjusts its predictions accordingly, providing a closed-loop system where past performance informs future campaign optimizations through automated recommendations.

Inventive Principle:

Principle #23Feedback

2Productivity

If content providers change campaign factors without knowing the actual causes of low delivery, then they can attempt to improve delivery, but the changes result in no improvement or little improvement

Engineering Contradiction:

Improvecontent delivery volumeVSAvoidlack of causal understanding

Core Design Contradiction:

ProductivityVSLoss of information

Solution Approach 1:

The patent introduces a machine learning model as an intermediary between raw campaign data and actionable insights. The gradient boosting model processes complex historical data, identifies causal relationships and important features, and translates them into understandable recommendations, bridging the gap between data and decision-making.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system analyzes changes in campaign parameters and their impact on delivery metrics by examining historical data. The machine learning model identifies which parameter changes are most likely to improve performance by learning from past campaigns, enabling targeted adjustments based on data-driven insights rather than random changes.

Inventive Principle:

Principle #35Parameter changes

3Quantity of substance

If content providers want to reach more users with their content, then they need to improve campaign delivery, but they lack the ability to determine which factors to change effectively

Engineering Contradiction:

Improvenumber of users reachedVSAvoiddifficulty in optimizing campaigns

Core Design Contradiction:

Quantity of substanceVSEase of operation

Solution Approach 1:

The patent implements self-service by enabling the system to automatically generate optimization recommendations without requiring expert manual analysis. The machine learning model autonomously processes campaign data, identifies optimization opportunities, and provides actionable recommendations, making the optimization process accessible to content providers regardless of their technical expertise.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system performs preliminary analysis by pre-processing and analyzing historical campaign data to build predictive models before new campaigns are launched. This advance preparation enables the system to provide immediate, data-driven recommendations for upcoming campaigns, eliminating the need for reactive adjustments after campaigns have already underperformed.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS10540683B2Machine-learned recommender system for performance optimization of network-transferred electronic content items

Publication Date: 2020.01.21 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

Machine learning techniques are described for generating recommendations using decision trees. A decision tree is generated based on training data that comprises multiple training instances, each of which comprises a feature value for each of multiple features and a label of a target variable. The multiple features correspond to attributes of multiple content delivery campaigns. Later, feature values of a content delivery campaign are received. The decision tree is traversed using the feature values to generate output. Based on the output, one or more recommendations are identified and the one or more recommendations are presented on a computing device.