Preface

摘要

This special issue contains four selected papers from the conference titled The Second International Workshop on Similarity Search and Applications (SISAP) and held in Prague in August, 2009. These contributions have been significantly extended and cover the theoretical as well as practical issues of similarity searching in metric spaces. The first manuscript introduces a novel problem of metric filtering, whose main difference from the traditional problem of searching in metric spaces is the fact that every data point carries its own personal distance to measure dissimilarity with queries. This may be helpful in several cases of recommendation systems, where user profiles are used to suggest new items to users. The second paper describes a new indexing structure based on locality-sensitive hashing principles. It maps metric space objects into a numeric domain and uses regular B-tree to speed up the search. The third contribution deals with indexing structures too and extends the GNAT structure, originally proposed by Sergey Brin. The last, fourth, manuscript studies the issues of distance functions applied to tree structures.