A wide range of criminal activities, particularly those related to hate crime, radicalisation, and terrorism, have an ever-increasing online component that needs to be considered, captured and investigated. Activity on the Dark Web is becoming more commonplace and can be a significant factor in investigations. Such information may need to be coupled with other open source information from the surface web and social media. Due to this, law enforcement is undergoing radical changes in their investigatory processes requiring new digital systems; however, aspects of the Dark Web still remain 'off-limits' to investigators as they battle technical, regulatory and organisational challenges. TENSOR is conceived as a system that can address many of these challenges by providing a mechanism for investigators to capture intelligence from online sources including the Dark Web and convert it into structured and coherent data. Within the system, TENSOR implements a unique data model, acquisition and extraction process that enables information to become searchable, analyse-able and relatable. Several analytical capabilities are layered above delivering advanced functionalities including natural language processing and concept extraction, machine translation, image recognition and iconography detection, social network analysis, stylometry, content similarity recommendations and image forensics. Each capability is presented through a unified intelligence dashboard that provides a clear entry point and logical analysis pathway for investigation to ensure TENSOR delivers a system that enables law enforcement to access and analyse complex information from the surface, deep and Dark Web.