Similarity Search Problem Research on Multi-dimensional Data Sets

摘要

In this paper, we present our continuous work on designing an algorithm to find nearest neighbors to given queries. In our previous work, we analyze the situation that there are multiple queries with different level of importance, and define a weight for each query point. We also propose an algorithm to find nearest neighbors to multiple queries with weights and enhanced our algorithm based on query point distribution. In this paper we analyze the data distribution on various dimensions, and apply the shrinking concept for the improvement and enhancement of our multi-query search approach.