PROJECT TITLE :

Effective and Efficient Algorithms for Flexible Aggregate Similarity Search in High Dimensional Spaces

ABSTRACT:

Varied applications in several fields, like spatial databases, multimedia databases, knowledge mining, and recommender systems, may benefit from efficient and effective combination similarity search, also called combination nearest neighbor (AggNN) search. Given a cluster of question objects , the goal of AggNN is to retrieve the most similar objects from the database, where the underlying similarity live is defined as an aggregation (typically add or max) of the distances between the retrieved objects and every query object in . Recently, the matter was generalized therefore as to retrieve the objects that are most just like a mounted proportion of the elements of . This variant of mixture similarity search is called “versatile AggNN”, or FANN. In this work, we have a tendency to propose two approximation algorithms, one for the add variant of FANN, and the other for the max variant. Intensive experiments are provided showing that, relative to state-of-the-art approaches (both actual and approximate), our algorithms turn out query results with sensible accuracy, while at the identical time being terribly efficient.