A knowledge graph (KG) consists of numerous triples, in which each triple, i.e., (head entity, relation, tail entity), denotes an assertion. By embedding KGs into low-dimensional representations, we could integrate them into deep learning models to perform various tasks, such as search and recommendation.

1.1 Question Answering over Knowledge Graphs
HamQA_TheWebConf23
KEQA_WSDM19

1.2 Knowledge Graph Refinement
Real-world KGs often contain many false assertions, which were inevitably injected during the construction of KGs. We propose to investigate error-aware KG learning to eliminate the impact of errors on KG-based systems.
KAEL_WSDM23
CAGED_CIKM22
LLM4EA_NeurIPS24

Networks are widely adopted to represent the relations between objects in many disciplines. In real-world scenarios, nodes are often associated with a rich set of data describing their characteristics. We model these systems as attributed networks. Our goal is to develop effective, scalable, and human-centric learning algorithms for attributed networks, to enable their actionable patterns to be easily accessible and interpretable to data consumers.

2.1 Datasets
BlogCatalog is an undirected network with 5,196 nodes and 171,743 edges, associated with node attributes of dimension 8,189.
Flickr is an undirected network with 7,575 nodes and 239,738 edges, associated with node attributes of dimension 12,047.
Yelp_multilabel is an undirected network with 249,012 nodes and 1,779,803 edges, associated with node attributes of dimension 20,000.

2.2 Attributed network embedding
FeatWalk_AAAI19
GraphRNA_KDD19
AANE_SDM17