Year: 2025
Type: article
Abstract: Approximate query on knowledge graphs (KGs) is an important and common task in real-world applications, where the goal is to return more results on KGs that match the query criteria. Previous approximate query methods have focused on static KGs. However, many KGs in real-world applications are dynamic and evolve over time. In this paper, we consider approximate queries in temporal knowledge graphs (TKGs) that may have specific timestamps in the predicates. We propose a Two-Level Approximate Query method (TLAQ) for temporal knowledge graphs based on the two-level embedding of vertex and graph. Specifically, we first improve the eigenmatrix of the GCN to enhance the embedding representation. On this basis, TLAQ defines relational reliability and attributive confidence at the vertex level. Then, we unify the encoding format of timestamps at the graph level to further strengthen the embedding model. Finally, we demonstrate the effectiveness of our proposed approach through a comprehensive experiment.
Source: Entropy
DOI: 10.3390/e27121232
Authors: Jiaxuan Liu, Xinyi Duan, Luyi Bai
Institution: Northeastern University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,137
Awards: G408208615, G5016398794, G7087789962
OpenAlex ID: W4417033785