Federica Baccini (University of Florence)
Nov 18, 2020 – 11:00 – 11:45 AM
Conference Meeting
Description
In this talk, I will introduce the problem of processing heterogeneous graphs (HETGs), i.e. graphs where nodes represent different entities, and edges denote multiple relation types. Despite heterogeneous graphs are able to model a broad variety of real world applications, only few state-of-the-art methods for performing usual tasks on homogeneous graphs (e.g. clustering, link prediction, node classification…) have been extended to the heterogeneous case. In particular, Neural Networks and Deep Learning have been shown to be efficient in processing graph structured data. Although most graph neural networks are designed to process homogeneous graphs, there are recent works that try to exploit these architectures to gather information from heterogeneous networks. The problems encountered in designing a model which efficiently extracts meaningful information from HETGs mainly regard difficulties in aggregating neighbourhood information and in processing heterogeneous node features.
In this seminar, I will first introduce the problem of HETGs processing, together with some examples of existing techniques for mining them. Then, I will give a motivation for the use of neural networks for processing HETGs, followed by some further perspectives and by a couple of possible applications.
Main References
Ziniu Hu et al.. Heterogeneous graph transformer. In: Proceedings of The Web Conference (2020).
Yizhou Sun et al.. Pathsim: Meta path-based top-k similarity search in heterogeneous information networks. In: Proceedings of the VLDB Endowment (2011).
Xiao Wang et al.. Heterogeneous graph attention network. In: The World Wide Web Conference (2019).
Zonghan Wu et al. A comprehensive survey on graph neural networks. In: IEEE Transactions on Neural Networks and Learning Systems (2020).
Chuxu Zhang et al.. Heterogeneous graph neural network. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (2019).