person behind the computer? Which content on the news Web portal is attracting the most attention? Which navigation pattern would lead readers to other news related to that content? Do trends in medical records indicate any new disease spreading in a given part of the world? Where are all my friends meeting? Can we detect any intra-day correlation clusters among stock exchanges? What are the top 10 emerging topics under discussion in the blogosphere, and who is driving the discussions?Although the information required to answer these questions is becoming increasingly available on the (Semantic) Web, there's currently no software system capable of computing the answersindeed, no system even lets users issue such queries. The reason is straightforward: answering such queries requires systems that can manage rapidly changing worlds at the semantic level.Of course, rapidly changing data can be analyzed on the fl y by specialized data-stream management systems, but such systems can't perform complex reasoning tasks, and they lack a protocol to publish widely and to provide access to the rapidly changing data.Reasoners, on the other hand, can perform such complex reasoning tasks, and the Semantic Web is providing the tools and methods to publish data widely on the Web. These technologies, however, don't really manage changing worlds: accessing and reasoning with rapidly changing information have been neglected or forgotten by their development communities.The state of the art in reasoning over changing worlds is based on temporal logic and belief revision; these are heavyweight tools, suitable for data that changes in low volumes at low frequency. Similarly, the problem of changing vocabularies and evolving ontologies has undergone thorough investigation, but here the standard practice relies on confi guration management techniques taken from software engineering, such as vocabulary and ontology versioning. These are suitable for ontologies that change on a weekly or monthly basis, but not for high-change-rate, high-frequency domains.Moreover, the typical (Semantic) Web architecture, which caches all the information, can hardly be applied to rapidly changing information, because the crawled data would be obsolete at the time of querying.We therefore suggest a completely different approach. Stream reasoning, an unexplored yet highimpact research area, is a new multidisciplinary approach that can provide the abstractions, foundations, methods, and tools required to integrate data streams, the Semantic Web, and reasoning systems, thus providing a way to answer our initial questions and many others.In this column, we describe two concrete examples of stream-reasoning applications and introduce stream-reasoning research problems. We also present a list of research areas that we believe should be investigated to turn stream reasoning into a reality.