An Access Structure for Similarity Search in Metric Spaces

Similarity retrieval is an important paradigm for searching in environments where exact match has little meaning Moreover, in order to enlarge the set of data types for which the similarity search can efficiently be performed, the mathematical notion of metric space provides a useful abstraction of similarity In this paper, we present a novel access structure for similarity search in arbitrary metric spaces, called D-Index D-Index supports easy insertions and deletions and bounded search costs for range queries with radius up to ρ D-Index also supports disk memories, thus, it is able to deal with large archives However, the partitioning principles employed in the D-Index are not very optimal since they produce high number of empty partitions We propose several strategies of partitioning and, finally, compare them.