Information Processing System for Dynamic Relevance Scoring

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing content-based filtering technologies rely on human intuition for parameter setting, leading to inaccurate identification of high-relevance content, while collaborative filtering struggles with content that is not frequently viewed due to limited information.

Innovation Solution

An information processing system that identifies data items with a high degree of action, acquires relevance information from user action histories, and sets parameters based on both the identified items and their details to enhance relevance scoring.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If content-based filtering with human intuition parameter setting is used, then the system can process any content including rarely viewed content, but the accuracy of identifying high-relevance content is low

Engineering Contradiction:

Improveaccuracy of identifying high-relevance contentVSAvoidcomplexity of parameter setting

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system automatically learns optimal parameters by analyzing user action histories and generating training data, eliminating the need for manual parameter setting by human experts. The algorithm self-adjusts parameters based on observed user behavior patterns, transforming parameter setting from a manual expert task to an automated learning process.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system uses user action histories as feedback to continuously improve parameter settings. By analyzing actual user interactions with content, the system refines its parameters to better predict relevance, creating a closed-loop system where performance is continuously optimized based on real-world usage data.

Inventive Principle:

Principle #23Feedback

2Measurement precision

If collaborative filtering based on user action history is used, then the accuracy of identifying relevant content is improved for frequently viewed content, but the system fails for rarely viewed content due to insufficient information

Engineering Contradiction:

Improveaccuracy of relevance identificationVSAvoidamount of action history information

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The system performs preliminary analysis of user action histories to identify frequently viewed content patterns before processing queries. By pre-processing and storing insights from popular content, the system prepares reference data that can be quickly applied even when specific content has limited interaction data, effectively pre-computing relevance signals.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system dynamically adjusts parameters based on the amount of available action history information. For content with sufficient interaction data, collaborative filtering parameters are optimized; for rarely viewed content, the system adapts parameters to rely more on content-based features and learned patterns from similar content, flexibly changing parameter weights based on data availability.

Inventive Principle:

Principle #35Parameter changes

3Ease of manufacture

If manual parameter setting based on human intuition is used, then the implementation is simple, but the accuracy of relevance scoring is insufficient

Engineering Contradiction:

Improveease of system implementationVSAvoidaccuracy of relevance scoring

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent replaces manual expert judgment (mechanical human decision-making) with automated machine learning algorithms. The system uses computational models to analyze user behavior patterns and determine optimal parameters, substituting human intuition with data-driven automated reasoning that scales without additional human effort.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system automatically generates training data from user action histories and performs self-training to optimize parameters. This self-service approach eliminates the need for manual parameter setting while maintaining simplicity of deployment, as the system configures itself upon receiving user interaction data.

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS11487835B2Information processing system, information processing method, and program

Publication Date: 2022.11.01 RAKUTEN GROUP INC

AI summary

Identification means of an information processing system identifies, based on an action history of each of a plurality of users, a plurality of data items having a degree of action equal to or more than a predetermined degree. First information acquisition means acquires, based on the action history of each user, first information indicating relevance between the respective data items identified by the identification means. Output means outputs, when details of each of the plurality of data items are input, second information indicating relevance between the respective data items based on a predetermined parameter. Setting means sets the parameter based on the details of each data item identified by the identification means and the first information.