Research fronts are the forefront of a scientific field. Timely identifying emerging trends and abrupt changes in scientific literature is not only beneficial for users of digital libraries, but also instrumental for users to trace the movement of a research front. A research front can be seen as a dynamic and transient mapping from the state of the art to its immediate prior art - the intellectual base. Citation and co-citation analysis primarily focuses on the structure and dynamics of the intellectual base, information retrieval and related research mainly focuses on intrinsic properties of text documents. In this article, we describe a novel approach that enables us to detect and visualize the transient relationship over time. Temporal associations between content words connected to a surge of interest in scientific papers and high-impact articles in the intellectual base are identified and visualized as a time-variant network. We apply the approach to the detection of thematic changes over time in information visualization and research in terrorism.