Similarity Search

Basic data for this project

Type of project: Own resources project
Duration: since 01/04/2011

Description

Similarity is a crucial paradigm for big data. It formalizes a conceptual relation among data objects along different abstraction levels in order to facilitate content-based data access and information analysis. The process of searching the most similar data objects is denoted as similarity search, which is considered to be a core operation for many supervised and unsupervised retrieval and analysis algorithms. Similarity search requires a formal definition of an appropriate similarity model, which describes a flexible means of data representation and a way of quantifying similarity between these data representations. While the data representation is responsible for modeling data objects and their inherent characteristics, the comparison between different representations is frequently established by means of similarity and dissimilarity measures. The more flexible the similarity model, the higher the adaptability to various challenges arising in different scientific and non-scientific domains. The objective of this research project is to develop and investigate similarity search concepts and techniques that scale to different domains.

Keywords: Similarity Search