An Efficient Indexing Method for Dynamic Graph kNN

k-nearest neighbor (kNN) search is a fundamental problem in graph mining. This search finds the k most relevant nodes to a given query node. The increased use of social network services and map applications due to the proliferation of mobile devices has necessitated faster searches. Although pre-constructing an index using graphs can accelerate a kNN search, existing methods struggle handling dynamic graph updates. Herein we propose an efficient index update method for dynamic graphs that utilizes a core-tree structure to efficiently update the index in response to dynamic changes in the graph. Our experimental analysis using real-world data demonstrated that the proposed method can construct indexes more efficiently than the state-of-the-art method.