Today there are massive volumes of text data, including for example news reports, research papers, and social media. Geospatial knowledge hypercube is defined as a multi-scale structure for integrating text data with heterogeneous geospatial data to discover latent connections and relationships through combining a variety of weakly supervised machine learning approaches. The hypercube lays a foundation for many knowledge discovery applications, such as recognizing geospatial entities and inferring geospatial relationships.
