Towards knowledge graph construction from entity co-occurrence
Nicolas Heist · MADOC (University of Mannheim) · 2018
The construction of knowledge graphs from resources on the Web is a topic that gained a lot of attention in recent years, especially with the uprising of large-scale cross-domain knowledge graphs like DBpedia and YAGO.Their successful exploitation of Wikipedia's structural elements like infoboxes and categories gives way to the thought that there is still a huge potential for approaches focused on structural elements of web documents.In this work, we present our research idea towards further exploitation of semi-structured data with a focus on entity cooccurrence.We want to explore the potential of co-occurrence patterns in varying contexts and test their generality when applying them to the Document Web.An overview of the state of the art is given and we show how our three-phased approach for the extraction of co-occurrence patterns fits in.Two planned experiments for the construction of knowledge graphs based on Wikipedia and the Document Web are sketched.Finally, potentials and limitations of the approach are discussed.