An Efficient Keyword Search on Temporal Graphs

Keyword search on temporal graph is to find a tree covering a set of query labels and being valid in the query time interval. It has many applications in cloud computing, community detection, social network, collaborative project, and so on. However, the existing methods are limited in solving the problem of keyword search on temporal graphs. We propose two basic algorithms, the discrete timestamp algorithm, and the approximate algorithm , the idea of which is trying to turn the problem into the traditional keyword search on graph. To address the low efficiency and low quality of the two basic algorithms, we propose a new algorithm based on dynamic programming to solve the keyword search on temporal graph. The idea is to extend a leaf vertex into a solution by edge-growth operation and tree-merger operation. We also propose two powerful pruning techniques to reduce the intermediate results during the extension. To speed up the algorithm further, we show that our algorithms can be easily parallel. Besides, all algorithms we proposed can handle different kinds of ranking functions well, and all of them can be adaptive to top-N keyword querying by simply changing the stop condition. The efficiency and effectiveness of the proposed algorithms are verified through extensive empirical studies.