TY  - JOUR
A1  - Bläsius, Thomas
A1  - Friedrich, Tobias
A1  - Lischeid, Julius
A1  - Meeks, Kitty
A1  - Schirneck, Friedrich Martin
T1  - Efficiently enumerating hitting sets of hypergraphs arising in data profiling
T2  - Journal of computer and system sciences : JCSS
N2  - The transversal hypergraph problem asks to enumerate the minimal hitting sets of a hypergraph. If the solutions have bounded size, Eiter and Gottlob [SICOMP'95] gave an algorithm running in output-polynomial time, but whose space requirement also scales with the output. We improve this to polynomial delay and space. Central to our approach is the extension problem, deciding for a set X of vertices whether it is contained in any minimal hitting set. We show that this is one of the first natural problems to be W[3]-complete. We give an algorithm for the extension problem running in time O(m(vertical bar X vertical bar+1) n) and prove a SETH-lower bound showing that this is close to optimal. We apply our enumeration method to the discovery problem of minimal unique column combinations from data profiling. Our empirical evaluation suggests that the algorithm outperforms its worst-case guarantees on hypergraphs stemming from real-world databases.
KW  - Data profiling
KW  - Enumeration algorithm
KW  - Minimal hitting set
KW  - Transversal hypergraph
KW  - Unique column combination
KW  - W[3]-Completeness
Y1  - 2022
UR  - https://publishup.uni-potsdam.de/frontdoor/index/index/docId/63442
SN  - 0022-0000
SN  - 1090-2724
VL  - 124
SP  - 192
EP  - 213
PB  - Elsevier
CY  - San Diego
ER  -