Accurate and Fast Retrieval for Complex Non-metric Data via Neighborhood Graphs

TL;DR

This paper introduces a graph-based search algorithm capable of efficiently handling complex non-metric and non-symmetric data directly, avoiding performance issues caused by traditional metric transformations.

Contribution

It demonstrates that constructing neighborhood graphs with modified distances enhances search performance for challenging data types, bypassing the need for metric symmetrization.

Findings

Graph-based search works with non-metric, non-symmetric data

Modified distance functions improve indexing performance

Avoids degradation caused by metric-space mapping

Abstract

We demonstrate that a graph-based search algorithm-relying on the construction of an approximate neighborhood graph-can directly work with challenging non-metric and/or non-symmetric distances without resorting to metric-space mapping and/or distance symmetrization, which, in turn, lead to substantial performance degradation. Although the straightforward metrization and symmetrization is usually ineffective, we find that constructing an index using a modified, e.g., symmetrized, distance can improve performance. This observation paves a way to a new line of research of designing index-specific graph-construction distance functions.