Multi-criteria Recommender Graph Expansion for High-Dimensional User Preferences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional recommender systems struggle to accurately recommend items based on multi-criteria evaluations, as they fail to effectively consider high-dimensional relationships between users and multiple criteria, leading to lower user satisfaction.
Innovation Solution
A multi-criteria recommender apparatus and method that expands item nodes into sub-nodes based on multiple evaluation criteria, using a neural network to generate a multi-criteria extended graph, where user and item nodes are connected with edges weighted by evaluation scores, allowing for personalized recommendations based on user preferences across various criteria.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a conventional single-criteria recommender apparatus uses a simple graph composed of user-item nodes, then the device complexity is low and ease of manufacture is improved, but the measurement precision of high-dimensional relationships between users and multiple criteria deteriorates
Solution Approach 1:
The patent segments each item node into multiple criterion-specific sub-nodes (e.g., quality sub-node, price sub-node, service sub-node). This segmentation allows the graph to represent multi-criteria relationships explicitly, improving the measurement precision of high-dimensional relationships between users and multiple criteria while maintaining a manageable graph structure through systematic division.
Solution Approach 2:
The patent transitions from a two-dimensional user-item graph to a multi-dimensional user-item-criterion graph by adding criterion dimensions. Each item is expanded into multiple sub-nodes representing different evaluation criteria, and edges are added to connect users to these sub-nodes based on their preferences, thereby capturing high-dimensional relationships in the recommendation system.
2Adaptability or versatility
If a multi-criteria recommender apparatus uses multiple graphs composed of user-item nodes according to different evaluation criteria, then the adaptability to diverse user needs is improved, but the device complexity increases
Solution Approach 1:
The patent merges multiple criterion-specific graphs into a single unified multi-criteria extended graph. Instead of maintaining separate graphs for quality, price, service, and other criteria, the invention integrates all criteria into one graph structure where item nodes are expanded into sub-nodes representing different criteria. This merging approach maintains adaptability to diverse user needs while reducing device complexity by eliminating the need to manage multiple separate graphs.
3Ease of operation
If a conventional recommender system uses a simple user-item graph, then the ease of operation is improved, but the recommendation accuracy for personalized multi-criteria preferences deteriorates
Solution Approach 1:
The patent applies local quality by assigning different weights to edges connecting users to different criterion sub-nodes based on user preferences. Each user can have customized weights for different criteria (e.g., user A values quality highly while user B values price highly). This local differentiation of edge weights maintains ease of operation through a unified graph structure while significantly improving recommendation accuracy by reflecting personalized multi-criteria preferences.
Data Source
AI summary
Provided are a multi-criteria recommender apparatus and method. The multi-criteria recommender apparatus obtains authorization for multi-criteria evaluation data provided by a user evaluating each item according to a plurality of different evaluation criteria to acquire a multi-criteria extended graph including a user node and an item node, wherein the item node is expanded into a plurality of sub-nodes according to the plurality of evaluation criteria, and selects a recommended item in consideration of the user's preferences for the plurality of evaluation criteria on the basis of embedding data that is obtained by performing a neural network operation on the multi-criteria extended graph.