Dynamic heterogeneous graph embedding via heterogeneous Hawkes process

Graph embedding, aiming to learn low-dimensional representations of nodes while preserving valuable structure information, has played a key role in graph analysis and inference. However, most existing methods deal with static homogeneous topologies, while graphs in real-world scenarios are gradually...

Full description

Saved in:
Bibliographic Details
Main Authors: JI, Yugang, JIA, Tianrui, FANG, Yuan, SHI, Chuan
Format: text
Language:English
Published: Institutional Knowledge at Singapore Management University 2021
Subjects:
Online Access:https://ink.library.smu.edu.sg/sis_research/6877
https://ink.library.smu.edu.sg/context/sis_research/article/7880/viewcontent/Dynamic_Heterogeneous_Graph_Embedding_via_Heterogeneous_Hawkes_Process.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Singapore Management University
Language: English
Description
Summary:Graph embedding, aiming to learn low-dimensional representations of nodes while preserving valuable structure information, has played a key role in graph analysis and inference. However, most existing methods deal with static homogeneous topologies, while graphs in real-world scenarios are gradually generated with different-typed temporal events, containing abundant semantics and dynamics. Limited work has been done for embedding dynamic heterogeneous graphs since it is very challenging to model the complete formation process of heterogeneous events. In this paper, we propose a novel Heterogeneous Hawkes Process based dynamic Graph Embedding (HPGE) to handle this problem. HPGE effectively integrates the Hawkes process into graph embedding to capture the excitation of various historical events on the current type-wise events. Specifically, HPGE first designs a heterogeneous conditional intensity to model the base rate and temporal influence caused by heterogeneous historical events. Then the heterogeneous evolved attention mechanism is designed to determine the fine-grained excitation to different-typed current events. Besides, we deploy the temporal importance sampling strategy to sample representative events for efficient excitation propagation. Experimental results demonstrate that HPGE consistently outperforms the state-of-the-art alternatives.