Joining relations under discrete uncertainty

TL;DR

This paper introduces and compares various algorithms for joining uncertain relations, demonstrating how data features influence their performance and how uncertainty statistics can guide optimal algorithm selection.

Contribution

It presents alternative algorithms based on sorting, indexing, and intermediate tables, and shows how to select the most efficient one using uncertainty statistics.

Findings

Algorithms perform differently depending on data features.

Uncertainty statistics can guide optimal algorithm choice.

Experimental comparison highlights efficiency variations.

Abstract

In this paper we introduce and experimentally compare alternative algorithms to join uncertain relations. Different algorithms are based on specific principles, e.g., sorting, indexing, or building intermediate relational tables to apply traditional approaches. As a consequence their performance is affected by different features of the input data, and each algorithm is shown to be more efficient than the others in specific cases. In this way statistics explicitly representing the amount and kind of uncertainty in the input uncertain relations can be used to choose the most efficient algorithm.