Efficient and Progressive Group Steiner Tree Search Rong-Hua Liy, Lu Qinz, Jeffrey Xu Yu∗, and Rui Maoy y Guangdong Province Key Laboratory of Popular High Performance Computers, Shenzhen University, China zCentre for QCIS, FEIT, University of Technology, Sydney, Australia ∗The Chinese University of Hong Kong, Hong Kong {rhli,mao}@szu.edu.cn;
[email protected];
[email protected] ABSTRACT graph, where each node denotes a tuple and each edge represents a The Group Steiner Tree (GST) problem is a fundamental prob- foreign key reference between two tuples. Each edge is associated lem in database area that has been successfully applied to key- with a weight, representing the strength of the relationship between word search in relational databases and team search in social net- two tuples (nodes). The keyword search problem aims to find a set works. The state-of-the-art algorithm for the GST problem is a of connected nodes (tuples) that covers all the given keywords with parameterized dynamic programming (DP) algorithm, which find- minimum total weight of all induced edges. Clearly, the optimal s the optimal tree in O(3kn + 2k(n log n + m)) time, where k solution for such a keyword search problem is a minimum-weight is the number of given groups, m and n are the number of the connected tree that covers the given keywords, and therefore it is edges and nodes of the graph respectively. The major limitations an instance of the GST problem. Most existing keyword search of the parameterized DP algorithm are twofold: (i) it is intractable systems such as BANKS-I [3], BANKS-II [19], BLINKS [17], S- even for very small values of k (e.g., k = 8) in large graphs due TAR [20] and DPBF [8, 6] are based on approximate or exact GST to its exponential complexity, and (ii) it cannot generate a solu- search techniques.