English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Heterogeneous graphs are pervasive in practical scenarios, where each graphconsists of multiple types of nodes and edges. Representation learning onheterogeneous graphs aims to obtain low-dimensional node representations thatcould preserve both node attributes and relation information. However, most ofthe existing graph convolution approaches were designed for homogeneous graphs,and therefore cannot handle heterogeneous graphs. Some recent methods designedfor heterogeneous graphs are also faced with several issues, including theinsufficient utilization of heterogeneous properties, structural informationloss, and lack of interpretability. In this paper, we propose HGConv, a novelHeterogeneous Graph Convolution approach, to learn comprehensive noderepresentations on heterogeneous graphs with a hybrid micro/macro levelconvolutional operation. Different from existing methods, HGConv could performconvolutions on the intrinsic structure of heterogeneous graphs directly atboth micro and macro levels: A micro-level convolution to learn the importanceof nodes within the same relation, and a macro-level convolution to distinguishthe subtle difference across different relations. The hybrid strategy enablesHGConv to fully leverage heterogeneous information with properinterpretability. Moreover, a weighted residual connection is designed toaggregate both inherent attributes and neighbor information of the focal nodeadaptively. Extensive experiments on various tasks demonstrate not only thesuperiority of HGConv over existing methods, but also the intuitiveinterpretability of our approach for graph analysis.

Hybrid Micro/Macro Level Convolution for Heterogeneous Graph Learning