This keynote will describe a number of projects being undertaken at the University of Queensland eResearch Lab that are pushing Semantic Web technologies to their limit to help solve grand challenges in the environmental, cultural and medical domains. In each of these use cases, we are integrating multi-modal data streams across space, time, disciplines, formats and agencies to infer and expose new knowledge through rich multi-layered and interactive visualizations. We are developing hypothesis-based query interfaces that provide evidence to validate or refute hypotheses and decision support services that recommend the optimum actions given current or predicted scenarios. We are using ontologies to influence and adapt government policies by linking policy-driven implementations, investments and management actions to real world indicators. Through evaluation of the methods and assessment of the achievements associated with the OzTrack, eReef, Skeletome and Twentieth Century in Paint projects, I will highlight those Semantic Web technologies that have worked for us and our user communities, those that haven't and those that need improvement. Finally, I will discuss what I believe will be the major outstanding research challenges facing Semantic Big Data in the next 5 years and those research areas with the greatest potential for impact.