10.1016/j.patrec.2023.02.004
https://zenodo.org/records/7940383
oai:zenodo.org:7940383
Mohamed, Hebatallah A.
Hebatallah A.
Mohamed
European Center for Living Technology (ECLT), Ca' Foscari University of Venice, Italy
Pilutti, Diego
Diego
Pilutti
European Center for Living Technology (ECLT), Ca' Foscari University of Venice, Italy
James, Stuart
Stuart
James
Pattern Analysis and Computer Vision (PAVIS), Istituto Italiano di Tecnologia (IIT), Italy
Del Bue, Alessio
Alessio
Del Bue
Pattern Analysis and Computer Vision (PAVIS), Istituto Italiano di Tecnologia (IIT), Italy
Pelillo, Marcello
Marcello
Pelillo
European Center for Living Technology (ECLT) & Department of Environmental Sciences, Informatics and Statistics (DAIS), Ca' Foscari University of Venice, Italy
Vascon, Sebastiano
Sebastiano
Vascon
European Center for Living Technology (ECLT) & Department of Environmental Sciences, Informatics and Statistics (DAIS), Ca' Foscari University of Venice, Italy
Locality-aware subgraphs for inductive link prediction in knowledge graphs
Zenodo
2023
Knowledge graphs
Inductive link prediction
Graph neural networks
Local clustering
Personalized PageRank
2023-02-15
eng
https://zenodo.org/communities/eu
Creative Commons Attribution 4.0 International
European Center for Living Technology (ECLT) & Department of Environmental Sciences, Informatics and Statistics (DAIS), Ca’ Foscari University of Venice, ItalyRecent methods for inductive reasoning on Knowledge Graphs (KGs) transform the link prediction problem into a graph classification task. They first extract a subgraph around each target link based on the
-hop neighborhood of the target entities, encode the subgraphs using a Graph Neural Network (GNN), then learn a function that maps subgraph structural patterns to link existence. Although these methods have witnessed great successes, increasing
often leads to an exponential expansion of the neighborhood, thereby degrading the GNN expressivity due to oversmoothing. In this paper, we formulate the subgraph extraction as a local clustering procedure that aims at sampling tightly-related subgraphs around the target links, based on a personalized PageRank (PPR) approach. Empirically, on three real-world KGs, we show that reasoning over subgraphs extracted by PPR-based local clustering can lead to a more accurate link prediction model than relying on neighbors within fixed hop distances. Furthermore, we investigate graph properties such as average clustering coefficient and node degree, and show that there is a relation between these and the performance of subgraph-based link prediction.
European Commission
10.13039/501100000780
870743
MEMEX: MEMories and EXperiences for inclusive digital storytelling