EFFECTIVE SIMILARITY SEARCH

Abstract

The data currently generated and collected increase fast not only in volume, but also in complexity, which brings great challenges to the field of data analysis. Similarity search has been acknowledged as one of the most important and useful query operators, and it plays a fundamental role of many applications. Thus, in this thesis, we investigate the problems concerning effective similarity search. We address the term "effective'' from three perspectives, aiming to provide our solutions to modern similarity search challenges. Firstly, we develop GENIE, a GPU-based Generic Inverted Index framework to support a wide variety of data types and similarity measures. Secondly, we propose a novel definition of diversity based on spatial angles for the problem of result diversification. Thirdly, we propose the personalized metric learning framework.