Keyword Graph: Answering Keyword Search over Large Graphs

In this paper, we focus on IR style queries, keyword search, over large disk-resident graphs. Since most existing approaches cache the whole graph and indexing structure in memory, these approaches cannot be applied into large graphs, such as RDF graphs and social networks. In this paper, we design a novel indexing structure,(kernel) keyword graph to summarize the structure of original graph. Based on (kernel) keyword graph, we propose an efficient keyword search algorithm. Extensive experiments confirm that our method can scale up to large graphs with millions of nodes and edges. The performance of our approach outperforms state-of-the-art algorithms by at least one order of magnitude.