Abstract |
We describe two success stories on the application of compact data structures (cds) to solve the problem of the excessively redundant space requirements posed by worst-case-optimal (wco) algorithms for multijoins in databases, and particularly basic graph patterns on graph databases. The aim of cds is to represent the data {\em and} additional data structures on it, using total space close to that of the plain (and, sometimes, compressed) data, while efficiently simulating the data structure operations. Cds turn out to be a perfect approach for the described problem: We designed and implemented cds that effectively use space close to that of the plain or compressed data, which is orders of magnitude less than existing systems, while retaining worst-case optimality and performing competitively with those systems in query time, sometimes being even considerably faster. |