4th Heterogeneous information network analysis and applications (HENA)
We are living in an increasingly interconnected world—where most of the data, informational objects, agents, or components are interconnected or interact with each other—forming gigantic and sophisticated information networks. Most real-world applications are grounded on heterogeneous information networks that include different types of objects or links, different from the conventional homogeneous networks with just a single type of objects or links. Previous works on heterogeneous information network analysis and its applications have led to a convergence of methodologies for network modeling, graph mining, and data semantics mining. The recent surge of graph embedding and graph neural networks also call for renewed and novel methodologies for heterogeneous information network analysis. As a promising network analysis paradigm, heterogeneous information network analysis also faces new challenges, such as how to manage more complex, high-order heterogonous semantics, how to efficiently handle large-scale heterogonous networks, and how to learn with limited labels on heterogeneous networks. Thus, the topics of the workshop is very well consistent with KDD, and are timely for researchers and practitioners to discuss emerging ideas in heterogeneous information network analysis.
Active research areas that are relevant to heterogeneous information networks include:
- Heterogeneous information network construction from complex data
- Data mining techniques on heterogeneous information networks, such as semantic mining, information diffusion and behavioral modeling, community detection and network evolution
- Graph representation learning for heterogeneous information networks
- Heterogeneous graph neural network
- Data mining based on knowledge graphs
- Parallel computing for heterogeneous information network analysis
- Applications of heterogeneous information network analysis in e-commerce, text mining, security, software engineering, etc
The emphasis of this workshop shall be analysis approaches and applications based on heterogeneous information networks extracted from heterogeneous sources such as technical literature, news articles, social network profile data, and social media. However, the scope is not limited to any particular approach to link analysis or any source of network information such as text corpora. Application areas that often exhibit a need for heterogeneous information network analysis include:
- Information diffusion and sharing systems: sensor networks, social media (opinions and sentiments, meme propagation, viral content, political commentary, etc.)
- Behavioral modeling: community recruitment and mass activity, large-scale patterns, traffic, spatiotemporal effects
- Content-management systems: version control, wikification
- Social recommender systems: communities, experts, friends, products, reviewers, providers
- Application areas: cybersecurity (information flow, trust networks, attack graphs, mechanism design), bioinformatics and biomedicine (genomics, proteomics, metabolomics), epidemiology
This workshop shall help to bring together people from these different areas and present an opportunity for researchers and practitioners to share new techniques for identifying and analyzing relationships in networks that integrate multiple types or sources of information.
Call for papers