Efficient negative sampling and retrieval fusion context path inductive relation prediction method

By employing semantic-aware negative sampling and context path extraction, efficient negative triples are generated and sentence embedding scores are applied. This addresses the efficiency and accuracy issues of knowledge graphs under dynamic changes, and improves the model's generalization ability and prediction performance of complex relationships.

CN119782551BActive Publication Date: 2025-11-04SHANDONG JIAOTONG UNIV
2 Cites 0 Cited by

Patent Information

Application Number
CN202411933293.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-12-26
Publication Date
2025-11-04
Estimated Expiration
2044-12-26

AI Technical Summary

Technical Problem

Existing knowledge graphs are inefficient and difficult to adapt to dynamic changes when new entities are added. Random negative sampling generates a lot of noise, and the understanding of path sparsity and complex relationships is insufficient, which affects the model's generalization ability and prediction accuracy.

Method used

We employ semantically aware negative sampling to generate negative triples, combine breadth-first search and recursive multi-hop context search to extract paths, generate detailed entity descriptions through a large language model and convert them into natural language sentences, and use the Sentence Transformer model for embedding and cosine similarity scoring.

Benefits of technology

It improves the prediction efficiency and accuracy of knowledge graphs, enhances the model's adaptability to dynamic changes, and improves the understanding and prediction performance of complex relationships.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application relates to an inductive relation prediction method of high-efficiency negative sampling and retrieval fusion context path, and belongs to the field of knowledge graph link prediction. The method comprises the following steps: semantic perception negative sampling; path extraction and filtering; context extraction and filtering; detailed information retrieval based on a large language model; sentence generation; embedding generation; and generation of a prediction result and training. The application generates more effective negative triples by introducing a semantic perception negative sampling technology, overcomes the limitations of traditional random negative sampling, extracts and filters paths and multi-hop contexts from query triples, combines long and short entities and relation descriptions into sentences, innovatively samples the multi-hop contexts of head entities and tail entities, and filters out irrelevant contents based on similarity, so that the problem of path scarcity or loss is effectively solved.
Need to check novelty before this filing date? Find Prior Art