The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Nos últimos anos, Redes Neurais de Grafos têm recebido enorme atenção da academia por seu enorme potencial de modelagem de características de rede, como macroestrutura e atributos de nó único. No entanto, os trabalhos anteriores concentram-se principalmente em redes homogêneas e não têm a capacidade de caracterizar a propriedade heterogênea da rede. Além disso, a maior parte da literatura anterior não consegue modelar a ligação de influência latente sob visão microscópica, tornando inviável modelar a relação conjunta entre a heterogeneidade e a interação mútua dentro do tipo de relação múltipla. Nesta carta, propomos uma estrutura de autoatenção baseada na influência latente para resolver as dificuldades mencionadas acima. Para modelar a heterogeneidade e as interações mútuas, redesenhamos o mecanismo de atenção com fator de influência latente no nível de relação de tipo único, que aprende o coeficiente de importância de seus vizinhos adjacentes sob os mesmos padrões baseados em meta-caminho. Para incorporar o meta-caminho heterogêneo em uma dimensão unificada, desenvolvemos uma nova estrutura baseada na autoatenção para a fusão da relação do meta-caminho de acordo com o coeficiente do meta-caminho aprendido. Nossos resultados experimentais demonstram que nossa estrutura não apenas alcança resultados mais elevados do que as atuais linhas de base do estado da arte, mas também mostra uma visão promissora na representação de relações interativas heterogêneas sob estruturas de rede complicadas.
Yang YAN
Tianjin University of Technology and Education
Qiuyan WANG
Tiangong University
Lin LIU
Tianjin LINHAITEC Ltd.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Yang YAN, Qiuyan WANG, Lin LIU, "Latent Influence Based Self-Attention Framework for Heterogeneous Network Embedding" in IEICE TRANSACTIONS on Information,
vol. E105-D, no. 7, pp. 1335-1339, July 2022, doi: 10.1587/transinf.2021EDL8093.
Abstract: In recent years, Graph Neural Networks has received enormous attention from academia for its huge potential of modeling the network traits such as macrostructure and single node attributes. However, prior mainstream works mainly focus on homogeneous network and lack the capacity to characterize the network heterogeneous property. Besides, most previous literature cannot the model latent influence link under microscope vision, making it infeasible to model the joint relation between the heterogeneity and mutual interaction within multiple relation type. In this letter, we propose a latent influence based self-attention framework to address the difficulties mentioned above. To model the heterogeneity and mutual interactions, we redesign the attention mechanism with latent influence factor on single-type relation level, which learns the importance coefficient from its adjacent neighbors under the same meta-path based patterns. To incorporate the heterogeneous meta-path in a unified dimension, we developed a novel self-attention based framework for meta-path relation fusion according to the learned meta-path coefficient. Our experimental results demonstrate that our framework not only achieves higher results than current state-of-the-art baselines, but also shows promising vision on depicting heterogeneous interactive relations under complicated network structure.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2021EDL8093/_p
Copiar
@ARTICLE{e105-d_7_1335,
author={Yang YAN, Qiuyan WANG, Lin LIU, },
journal={IEICE TRANSACTIONS on Information},
title={Latent Influence Based Self-Attention Framework for Heterogeneous Network Embedding},
year={2022},
volume={E105-D},
number={7},
pages={1335-1339},
abstract={In recent years, Graph Neural Networks has received enormous attention from academia for its huge potential of modeling the network traits such as macrostructure and single node attributes. However, prior mainstream works mainly focus on homogeneous network and lack the capacity to characterize the network heterogeneous property. Besides, most previous literature cannot the model latent influence link under microscope vision, making it infeasible to model the joint relation between the heterogeneity and mutual interaction within multiple relation type. In this letter, we propose a latent influence based self-attention framework to address the difficulties mentioned above. To model the heterogeneity and mutual interactions, we redesign the attention mechanism with latent influence factor on single-type relation level, which learns the importance coefficient from its adjacent neighbors under the same meta-path based patterns. To incorporate the heterogeneous meta-path in a unified dimension, we developed a novel self-attention based framework for meta-path relation fusion according to the learned meta-path coefficient. Our experimental results demonstrate that our framework not only achieves higher results than current state-of-the-art baselines, but also shows promising vision on depicting heterogeneous interactive relations under complicated network structure.},
keywords={},
doi={10.1587/transinf.2021EDL8093},
ISSN={1745-1361},
month={July},}
Copiar
TY - JOUR
TI - Latent Influence Based Self-Attention Framework for Heterogeneous Network Embedding
T2 - IEICE TRANSACTIONS on Information
SP - 1335
EP - 1339
AU - Yang YAN
AU - Qiuyan WANG
AU - Lin LIU
PY - 2022
DO - 10.1587/transinf.2021EDL8093
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E105-D
IS - 7
JA - IEICE TRANSACTIONS on Information
Y1 - July 2022
AB - In recent years, Graph Neural Networks has received enormous attention from academia for its huge potential of modeling the network traits such as macrostructure and single node attributes. However, prior mainstream works mainly focus on homogeneous network and lack the capacity to characterize the network heterogeneous property. Besides, most previous literature cannot the model latent influence link under microscope vision, making it infeasible to model the joint relation between the heterogeneity and mutual interaction within multiple relation type. In this letter, we propose a latent influence based self-attention framework to address the difficulties mentioned above. To model the heterogeneity and mutual interactions, we redesign the attention mechanism with latent influence factor on single-type relation level, which learns the importance coefficient from its adjacent neighbors under the same meta-path based patterns. To incorporate the heterogeneous meta-path in a unified dimension, we developed a novel self-attention based framework for meta-path relation fusion according to the learned meta-path coefficient. Our experimental results demonstrate that our framework not only achieves higher results than current state-of-the-art baselines, but also shows promising vision on depicting heterogeneous interactive relations under complicated network structure.
ER -