Aggregate Nearest Neighbor Queries

Speaker: Ken YIU

Abstract

Spatial data management applications (e.g., location-based services) require efficient processing of aggregate nearest neighbor (ANN) queries for query and data points. Consider for example, a set of users at specific locations (query points) that want to find the restaurant (data object) that minimizes the total effort for them to meet there. In addition, this problem is also important for clustering and outlier detection.

The talk will give a formal definition of aggregate nearest neighbor (ANN) queries and introduce algorithms for them. Previous work focuses on the sum function for objects in Euclidean space. We study this problem for any monotone aggregate functions and also extend this problem for objects on spatial (road) networks. Some experimental results will be presented.