Attention-Driven Metapath Encoding in Heterogeneous Graphs
This work addresses the problem of capturing semantic structures in heterogeneous graphs for researchers in graph machine learning, but it is incremental as it builds on existing aggregation mechanisms and training schedulers.
The paper tackles node classification in heterogeneous graphs by incorporating attention mechanisms to encode entire metapaths without dropping intermediate nodes, resulting in a competitive model that matches state-of-the-art performance on the IMDB benchmark dataset.
One of the emerging techniques in node classification in heterogeneous graphs is to restrict message aggregation to pre-defined, semantically meaningful structures called metapaths. This work is the first attempt to incorporate attention into the process of encoding entire metapaths without dropping intermediate nodes. In particular, we construct two encoders: the first uses sequential attention to extend the multi-hop message passing algorithm designed in \citet{magna} to the metapath setting, and the second incorporates direct attention to extract semantic relations in the metapath. The model then employs the intra-metapath and inter-metapath aggregation mechanisms of \citet{han}. We furthermore use the powerful training scheduler specialized for heterogeneous graphs that was developed in \citet{lts}, ensuring the model slowly learns how to classify the most difficult nodes. The result is a resilient, general-purpose framework for capturing semantic structures in heterogeneous graphs. In particular, we demonstrate that our model is competitive with state-of-the-art models on performing node classification on the IMDB dataset, a popular benchmark introduced in \citet{benchmark}.