Proceedings of the Third International Conference on SImilarity Search and APplications (SISAP 2010)

This volume contains 2 invited papers, 11 regular papers, 2 posters and 6 demos, presented at SISAP 2010 conference, held in Istanbul, Turkey, on September 18-19, 2010. The papers were selected from 29 submissions, by a renowned program committee. We received submissions from all over the world (Argentina, Chile, China, Cuba, Czech Republic, Germany, Greece, Italy, Japan, Mexico, Norway, Spain, and USA). In this third volume of the SISAP series, the acceptance ratio of 66% was achieved as a result of the increasing quality of submissions established last year. The papers were selected based on their originality, relevance, and technical strength. The submissions cover the spectrum of similarity searching topics from novel indexing techniques, to approximate methods for speeding up similarity search in metric/non-metric spaces, to applications in computerized biology, image classifications, and content-based image retrieval. In the demonstrations section a large number of interesting prototypes of similarity search engines, related to content- and semantics-based retrieval of multimedia data (audio, images, and video), was presented. Finally, the posters present an approach for balancing splits in indexing techniques based on generalized hyperplane partitioning and a system for automatic video annotation based on similarity among detected shots. Two invited speakers gave excellent and thought-provoking talks. Prof. Vladimir Pestov provided interesting connections between similarity searching in metric spaces, the phenomenon of concentration of measure on high-dimensional structures, and the Vapnik-Chervonenkis theory of statistical learning, deriving bounds on performance of well-know indexing structures and showing how statistical learning methods can be used to build indexing schemes in datasets drawn from distributions of lower intrinsic dimension. Prof. Tomas Skopal discussed the impact of the metric indexing paradigm on real-world applications, analyzing the world of similarity searching in metric spaces from the point of view of practitioners, transforming his conclusions into research challenges and perspectives of the similarity search methods. These contributions given by leading researchers definitely contributed to an in-depth and up-to-date understanding of fascinating fields and will surely stimulate further research. Extended versions of the best SISAP 2010 papers were collected and will be published in a special issue of Journal of Discrete Algorithms (Elsevier). We are grateful to the program committee members, the local organizers, and the external reviewers for their committed and high-quality work. We highly appreciate the work done by Karina Figueroa for maintaining the Metric Library, and those who contributed to it. Reviewing of papers and preparation of the proceedings were done using the EasyChair conference system (www.easychair.org), for which we specially thank Andrei Voronkov. Last but not least, we thank Adrienne Griscti (ACM program manager for SIG publications) for her hard work in helping us publishing these proceedings.