SPOT databases have been proposed as a paradigm for efficiently reasoning about probabilistic spatio-temporal data. A selection query asks for all pairs of objects and times such that the object is within a query region with a probability within a stated probability interval. Two alternative semantics have been introduced for selection queries: optimistic and cautious selection. It has been shown in past work that selection is characterized by a linear program whose solutions correspond to certain kinds of probability density functions (pdfs). In this chapter, we define a space called the SPOT PDF Space (SPS for short) and show that the space of solutions to a cautious selection query is a convex polytope in this space. This convex polytope can be approximated both by an interior region and a containing region. We show that both notions can be jointly used to prune the search space when answering a query. We report on experiments showing that cautious selection can be executed in about 4 seconds on databases containing 3 million SPOT atoms. © 2010 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Parisi, F., Parker, A., Grant, J., & Subrahmanian, V. S. (2010). Scaling cautious selection in spatial probabilistic temporal databases. Studies in Fuzziness and Soft Computing, 256, 307–340. https://doi.org/10.1007/978-3-642-14755-5_12
Mendeley helps you to discover research relevant for your work.