LSimRank: Node Similarity in a Labeled Graph.

The notion of node similarity is useful in many real-world applications. Many existing similarity measurements such as SimRank and its variants have been proposed. Among these measurements, most capture the structural information of a graph only, and thus they are not suitable for graphs with additional label information. We propose a new similarity measurement called LSimRank which measures the similarities among nodes by using both the structural information and the label information of a graph. Extensive experiments on datasets verify that LSimRank is superior over SimRank and other variants on labeled graphs.