A Learned Index for Exact Similarity Search in Metric Spaces

Indexing is an effective way to support efficient query processing in large databases. Recently the concept of learned index , which replaces or complements traditional index structures with machine learning models, has been actively explored reduce storage and search costs. However, accurate similarity high-dimensional metric spaces remains be open challenge. In this paper, we propose a novel indexing approach called LIMS that uses data clustering, pivot-based transformation techniques learned indexes spaces. LIMS, underlying partitioned into clusters such each cluster follows relatively uniform distribution. Data redistribution achieved by utilizing small number pivots for cluster. Similar are mapped compact regions values totally ordinal. Machine models developed approximate position record on disk. Efficient algorithms designed range queries nearest neighbor based maintenance dynamic updates. Extensive experiments real-world synthetic datasets demonstrate superiority compared state-of-the-art indexes.