Symmetric M-tree

TL;DR

This paper introduces a deletion algorithm for the M-tree, a balanced metric access method, enhancing its functionality while maintaining performance, thus enabling efficient object removal in metric databases.

Contribution

It proposes a novel Delete algorithm for the M-tree, complementing its Insert method and improving its operational capabilities.

Findings

The Delete algorithm is effective and maintains the tree's performance.

The modified M-tree supports both insertion and deletion operations efficiently.

Performance remains comparable to the original M-tree.

Abstract

The M-tree is a paged, dynamically balanced metric access method that responds gracefully to the insertion of new objects. To date, no algorithm has been published for the corresponding Delete operation. We believe this to be non-trivial because of the design of the M-tree's Insert algorithm. We propose a modification to Insert that overcomes this problem and give the corresponding Delete algorithm. The performance of the tree is comparable to the M-tree and offers additional benefits in terms of supported operations, which we briefly discuss.