Goods Knowledge Graph Training with Entity-Attribute Embedding Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for recommending goods based on user interests lack accuracy due to inadequate correlation between goods and entity nodes in knowledge graphs, resulting in low recommendation precision.
Innovation Solution
A method and device for training a goods knowledge graph using graph embedding models, incorporating both head entity-relation-tail entity and entity-attribute-value triples to obtain embedding vectors, and utilizing these vectors in graph convolution network models for personalized recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional knowledge graph methods are used for goods recommendation, then the system complexity is low, but the recommendation accuracy is insufficient
Solution Approach 1:
The patent merges entity triples and attribute triples into a unified knowledge graph structure, combining relational information with attribute information to enhance recommendation accuracy while maintaining manageable system complexity through integrated design
Solution Approach 2:
The patent introduces embedding vectors as an intermediary representation layer between the knowledge graph and the recommendation model, transforming structured knowledge into continuous vector space that improves recommendation precision without significantly increasing system complexity
2Reliability
If only entity-relation triples are used in the knowledge graph, then the construction process is simple, but the correlation between goods and entity nodes is insufficient
Solution Approach 1:
The patent combines entity-relation triples with entity-attribute triples into a unified knowledge graph, merging relational structure with attribute information to strengthen the correlation between goods and entity nodes while maintaining a coherent graph structure
Solution Approach 2:
The patent creates a composite knowledge graph structure that integrates different types of triples (entity-relation and entity-attribute), combining diverse information types to enhance the representational power and correlation strength of the knowledge graph
Data Source
AI summary
Provided are a method and device for recommending goods, a method and device for training a goods knowledge graph, and a method and device for training a model. The method for training a goods knowledge graph includes: constructing an initial goods knowledge graph based on a first type of triples and a second type of triples, where a format of the first type of triples is head entity-relation-tail entity, and a format of the second type of triples is entity-attribute-attribute value (S101); and training the initial goods knowledge graph based on a graph embedding model to obtain embedding vectors of entities in the trained goods knowledge graph (S102).