The global digital library development should provide not only the first-hand scientific literature needed by scientists, researchers and practitioners, but also value-added services that can go beyond the merely needs for information access. We describe a generic approach to tracking and visualising the structure of a knowledge domain. The goal is to summarise and synthesis latent structures of knowledge domains as reflected through scientific literature available on the Internet, in a wide range of digital libraries, and publishers' online outlets. This approach extends our work in visualising a subject domain based on author citation and co-citation patterns derived from citation databases. We illustrate how this approach can be used to capture the research front of a knowledge domain based on more volatile input sources such as dynamic queries to the Web of Science, which is a Web-based citation database. In order to demonstrate this approach, we include one example of visualising the knowledge structure concerning the extinction of dinosaurs. This example is used as a vehicle to elaborate important features of our approach. We highlight the potential of our approach and underlying information techniques in achieving the visualisation of knowledge structures.