For many geospatial and multimedia database applications, a fundamental challenge is to answer similaritysearch on large amounts of data, also known as the nearest neighbor (NN) query. Despite extensivestudies, little is known for aggregate similarity search that has witnessed an increasing number of applications.Given a data set P, an aggregate similarity query is specified by a group of query objects Q, anaggregator ? and a similarity function f involving at least the spatial distance between objects. The goalis to find an object p? ? P such that the aggregate similarity between p? and objects from Q is minimized.For example, when ? is sum over spatial distances, it is to find p? ? P that minimizes the sum of distancesbetween p? and every object from Q.The aggregate similarity search presents interesting and challenging research problems, in terms of bothquery semantics and query processing techniques. We will also explore parallel and relational (those thatcan be directly implemented by standard SQL statements) methods in this project.Due to the fundamental importance of the similarity search and the emerging applications for variousaggregate similarity search, our project will significantly improve the scientific work in geospatial intelligenceanalytics, including spatial intelligence, complex spatial data analysis, GIS, and location-based services.End-users may design customized aggregate similarity search to identify normal and discover anomalies incomplex geospatial and multimedia data.