DSACL+-tree: A Dynamic Data Structure for Similarity Search in Secondary Memory

摘要: Metric space searching is an emerging technique to address the problem of efficient similarity in many applications, including multimedia databases and other repositories handling complex objects. Although promising, metric approach still immature several aspects that are well established traditional databases. In particular, most indexing schemes not dynamic. From few dynamic indexes, even fewer work secondary memory. That is, them need index main memory order operate efficiently. this paper we introduce two different secondary-memory versions Dynamic Spatial Approximation Tree with Clusters (DSACL-tree from Barroso et al.) which has shown be competitive These indexes handle scenario state art. But particular innovations proposed by version DSACL+-tree lead significant performance improvements.The resulting data structures can useful a wide range database application.