Aspect-Enhanced Explainable Recommendations via Knowledge Graph
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional recommendation systems fail to provide accurate and explainable recommendations, especially in complex environments where users have diverse and specific interests, and struggle with cross-domain recommendations due to limited data sharing and reliance on surface-level explanations.
Innovation Solution
A system and method for generating aspect-enhanced explainable description-based recommendations by identifying aspects of items from text-based descriptions, extracting subjective knowledge, and representing this information in a knowledge graph to recommend items based on aspects, while protecting underlying data, allowing for cross-provider and cross-domain recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional recommendation systems use basic user and product information, then the system is simple to implement, but the recommendation accuracy and explainability are insufficient
Solution Approach 1:
The patent segments the recommendation system into multiple components: aspect extraction module, knowledge graph construction module, and recommendation generation module. Each module handles specific tasks (extracting aspects from text, building structured relationships, generating recommendations with explanations), allowing the complex system to be managed through modular functional divisions while improving overall recommendation accuracy.
Solution Approach 2:
The patent introduces a knowledge graph as an intermediary structure between user preferences and item recommendations. The knowledge graph contains nodes representing users, items, and aspects, with edges representing relationships and weights. This intermediary enables complex reasoning about user interests and item characteristics, improving recommendation accuracy and explainability without requiring direct complex processing between all user-item pairs.
2Loss of information
If recommendation systems rely on surface-level explanations, then the system is easier to operate, but the explanations are not sufficiently informative for users with diverse interests
Solution Approach 1:
The patent adds a new dimension to recommendations by incorporating aspect-based explanations. Instead of only providing item recommendations, the system extracts multiple aspects (features, attributes, characteristics) from item descriptions and presents them as structured explanations. This additional dimensional information provides users with comprehensive insights into why items are recommended, addressing diverse interests without complicating the core recommendation function.
3Adaptability or versatility
If companies share detailed datasets for cross-domain recommendations, then recommendation accuracy improves, but data security and privacy protection are compromised
Solution Approach 1:
The patent extracts only the necessary aspect information from detailed item descriptions and user data, rather than sharing complete datasets. The aspect extraction module identifies and extracts relevant features, attributes, and characteristics, converting comprehensive data into condensed aspect representations. This extraction approach enables cross-domain recommendations by sharing only essential information, improving adaptability while minimizing data security risks by not exposing detailed proprietary datasets.
Data Source
AI summary
A recommendation method includes determining one or more aspects of a first item based on at least one descriptive text of the first item. The recommendation method also includes updating a knowledge graph containing nodes that represent multiple items, multiple users, and multiple aspects. Updating the knowledge graph includes linking one or more nodes representing the one or more aspects of the first item to a node representing the first item with one or more first edges. Each of the one or more first edges identifies weights associated with (i) user sentiment about the associated aspect of the first item and (ii) an importance of the associated aspect to the first item. In addition, the recommendation method includes recommending a second item for a user with an explanation based on at least one aspect linked to the second item in the knowledge graph.