AI Group Recommendation Re-Ranking with Knowledge Graph Preferences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI systems struggle to make joint predictions for groups of entities while respecting individual preferences and relationships, leading to a loss of information and disregard for preference strengths.
Innovation Solution
A method and system that consolidates individual entity predictions using a knowledge graph with source, target, and attribute entities, incorporating preference weights and relationships to generate a joint probability-based recommendation with explanations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If individual entity predictions are made separately, then individual preferences can be captured, but information loss occurs and relationships between entities are ignored
Solution Approach 1:
The patent combines multiple individual entity predictions into a single joint prediction by merging the prediction results and re-ranking them based on group-level relationships. This merging process prevents information loss by integrating individual preferences while simultaneously capturing inter-entity relationships that would be missed in isolated predictions.
Solution Approach 2:
The prediction process is segmented into distinct stages: individual entity prediction, relationship-based re-ranking, and final joint prediction generation. This segmentation allows the system to handle complex relationships systematically while maintaining the benefits of individual predictions, thereby reducing overall system complexity through structured processing.
2Adaptability or versatility
If group prediction is made as a whole, then joint decision can be achieved, but individual preferences and relationship strengths are disregarded
Solution Approach 1:
The patent applies local quality by differentiating the treatment of individual entities within the group. Each entity's prediction is initially generated with its own specific characteristics and preferences, then re-ranked based on local relationship strengths with other group members. This ensures that individual preferences are preserved while adapting to group dynamics.
Solution Approach 2:
The system dynamically adjusts the ranking of predictions based on relationship strengths that are calculated for each group configuration. The re-ranking process is dynamic and adapts to the specific relationships present in each group, allowing the system to maintain high preference accuracy while achieving versatile group predictions.
3Ease of operation
If majority vote is used for consolidation, then simple joint decision can be reached, but preference strengths and individual influences are not respected
Solution Approach 1:
Instead of using a simple majority vote, the patent changes the consolidation parameter from equal weighting to relationship-strength-based weighting. The re-ranking process incorporates parameters that reflect the strength of relationships between entities, making the consolidation more reliable by considering preference strengths while maintaining operational simplicity through an automated re-ranking mechanism.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of consolidating recommendations based on individual recommendations includes receiving a knowledge graph including source entities, target entities and attribute entities. Each source entity and each target entity is linked to one or more of the attribute entities. Using a trained prediction learning model, a prediction is determined for each source entity based on the knowledge graph. The trained prediction model was trained using prediction training data including historical data. The prediction for each source includes recommendation data identifying one or multiple target entities. Using a trained consolidation learning model, a consolidated prediction is determined for the source entities based on the prediction for each source entity. The trained consolidation learning model was trained using consolidation training data including the historical data and the recommendation data. The consolidated prediction identifies a target entity that maximizes a joint probability of the source entities.