Japan Society for the Promotion of Science:Grants-in-Aid for Scientific Research Grant-in-Aid for Scientific Research (C)
Date (from‐to) : 2008 -2010
Author : UEDA Naonori; SAITO Kazumi; YAMADA Takeshi; SAWADA Hiroshi; AOYAMA Kazuo
This research project proposed a similarity search method that utilized a degree-reduced k-nearest neighbor graph (k-DR graph), which is a small-world network (or graph) in complex networks, for a search index structure, and that explored the k-DR graph by a greedy search or a best-first search algorithm. We have experimentally confirmed that this similarity search method was applicable to a variety of media types such as documents, images, and speech signals (utterances) and quickly found objects similar to a given query. The proposed similarity method with the foregoing properties is expected to serve as a base for novel similarity search techniques.