Top-K Correlation Sub-graph Search in Graph Databases

Recently, due to its wide applications, (similar) subgraph search has attracted a lot of attentions from database and data mining community, such as [13, 18, 19, 5]. In [8], Ke et al. first proposed correlation sub-graph search problem (CGSearch for short) to capture the underlying dependency between subgraphs in a graph database, that is CGS algorithm. However, CGS algorithm requires the specification of a minimum correlation threshold θ to perform computation. In practice, it may not be trivial for users to provide an appropriate threshold θ, since different graph databases typically have different characteristics. Therefore, we propose an alternative mining task: top-K correlation subgraph search(TOP-CGSearh for short). The new problem itself does not require setting a correlation threshold, which leads the previous proposed CGS algorithm inefficient if we apply it directly to TOP-CGSearch problem. To conduct TOPCGSearch efficiently, we develop a pattern-growth algorithm (that...

• Type: Conference
• Year: 2009
• Where: DASFAA
• Authors: Lei Zou, Lei Chen 0002, Yansheng Lu
• Keywords: Correlation Threshold, DASFAA 2009, Database, Minimum Correlation Threshold, first Proposed Correlation