Archaeoinformatics - Data Science

DAAD PPP with UIC: Heterogeneous Information Network Management and Analysis


In collaboration with Philip S. Yu, UIC (University of Illinois at Chicago).

Most real system consists of multi-typed entities/objects with a variety of relationships between interacting and/or associated objects. These interactions and relationships between entities are naturally represented as information network graphs. Information networks are ubiquitous and well-established in the real world with  diverse applications fields such as publication networks, communication networks, the World Wide Web, or social networks. Nowadays, we have to handle a size of these information networks with ranges from hundreds up to millions and billions of nodes. By the rise of data integration, an increasing attention to information networks can be observed in academia and industry. Related to this development, new challenges have been introduced as we are not only concentrating on homogeneous data, i.e. we only have one type of objects or relationships between the objects in our network, but are rather faced to heterogeneous data derived from a variety of sources. Information networks are often organized in form of Resource Description Framework (RDF) data. RDF provides a simple way for expressing facts across linked data and is appropriate for the representation of information networks and/or knowledge graphs. Several distributed and federated RDF systems have emerged to handle the massive amounts of available RDF data nowadays. However, the weak (or missing) structure of information networks or knowledge graphs makes the development of methods for the efficient organization of this data difficult. In this project, we consider methods for the efficient organization of Heterogeneous Information Networks (HINs). HINs provide not only a more general, natural, and rich representation of relationships between objects and semantic information than traditional (homogeneous) networks but also follow a certain graph schema that provides important information about the structure of the graph. Consequently, the problem of understanding the vast amount of information modeled in heterogeneous information networks has received a lot of interest. Though many approaches have been developed to efficiently handle RDF data in the context of general Knowledge Graphs, standard methods for the efficient organization of knowledge structured in form of Heterogeneous Information Networks and corresponding scalable methods for performing relationship queries are still subject to future work.

Research Objectives:

  • Investigation of methods for the efficient management of Heterogeneous Information Networks.
  • Investigation of scalable methods for meta-structure-based relationship queries and relationship pattern analysis.

For more details, please contact Prof. Dr. Matthias Renz, Christian Beth, M.Sc.