Databases and the semantic Web: Data semantics revisited

Abstract

Two panels, held at SIGMOD'98 (Seattle, June 4) and CAiSE' 98 (Pisa, June 11), discussed the topic of data semantics and its place in Databases research in the next millennium. The first, titled "Next Generation Database Systems Won't Work Without Semantics" included as panelists Philip Bernstein, Umesh Dayal, John Mylopoulos (chair), Sham Navathe and Marek Rusinkiewicz. The second one, titled "Data Semantics Can't Fail This Time!" included as panelists Michael Brodie, Stefano Ceri, John Mylopoulos (chair), and Arne Solvberg. Atypically for panels, participants to both discussions generally agreed that data semantics will be the problem for Databases researchers to tackle in the near future. Stefano Ceri summed up well the sentiments of the discussions by declaring that " . . . The three most important research problems in Databases used to be 'Performance', 'Performance', and 'Performance'; in years to come, the three most important and challenging problems-will be 'Semantics', 'Semantics', and 'Semantics'. . . " What is the data semantics problem? In what sense did it "fail" in the past? . . . And why did the experts agree - unanimously - that the situation was about to change? We review the data semantics problem and its long history in Databases research, noting the reasons why solutions of the past won't work in the future. We then review recent work on the Semantic Web and the directions it is taking. Finally, we sketch two new directions for research on data semantics. © Springer-Verlag Berlin Heidelberg 2005.