The aim of the project is to develop novel techniques to construct advanced types of Knowledge Graphs. Current generation of Knowledge Graphs is mostly limited to unary and binary factual knowledge; they do not have a particularly strong model for knowledge whose existence depends critically on certain spatial/temporal/contextual conditions.

The challenges involved in this research include:

  • How to model and extract spatial/temporal/contextual knowledge?
  • How to model and extract n-ary relation?
  • How to extract commonsense Knowledge?


Key Contact

Associate Professor Wei Wang ([email protected])


Syeed Ibn Faiz, Robert E. Mercer: Extracting Higher Order Relations From Biomedical Text. ACL 2014

Johannes Hoffart, Fabian M. Suchanek, Klaus Berberich, Gerhard Weikum: YAGO2: A spatially and temporally enhanced knowledge base from Wikipedia. Artif. Intell. 194: 28-61 (2013)

Jiaqiang Chen, Niket Tandon, Gerard de Melo: Neural Word Representations from Large-Scale Commonsense Knowledge. WI-IAT (1) 2015: 225-228

Daniel Hernández, Aidan Hogan, Markus Krötzsch: Reifying RDF: What Works Well With Wikidata? [email protected] 2015: 32-47