GenPR: Generative PageRank Framework for Semi-supervised Learning on Citation Graphs - Archive ouverte HAL Access content directly
Conference Papers Year : 2020

GenPR: Generative PageRank Framework for Semi-supervised Learning on Citation Graphs

(1, 2) , (1, 2)
1
2

Abstract

Nowadays, Semi-Supervised Learning (SSL) on citation graph data sets is a rapidly growing area of research. However, the recently proposed graph-based SSL algorithms use a default adjacency matrix with binary weights on edges (citations), that causes a loss of the nodes (papers) similarity information. In this work, therefore, we propose a framework focused on embedding PageRank SSL in a generative model. This framework allows one to do joint training of nodes latent space representation and label spreading through the reweighted adjacency matrix by node similarities in the latent space. We explain that a generative model can improve accuracy and reduce the number of iteration steps for PageRank SSL. Moreover, we show that our framework outperforms the best graph-based SSL algorithms on four public citation graph data sets and improves the interpretability of classification results.
Fichier principal
Vignette du fichier
paper_26-3(1).pdf (930.64 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-02977308 , version 1 (24-10-2020)

Identifiers

Cite

Mikhail Kamalov, Konstantin Avrachenkov. GenPR: Generative PageRank Framework for Semi-supervised Learning on Citation Graphs. INL 2020 - 9th Conference on Artificial Intelligence and Natural Language, Oct 2020, Helsinki, Finland. pp.158-165, ⟨10.1007/978-3-030-59082-6_12⟩. ⟨hal-02977308⟩
86 View
175 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More