Towards Large-scale Inconsistency Measurement

TL;DR

This paper introduces a new stream-based inconsistency measure for large knowledge bases, along with approximations for existing measures, demonstrating feasibility through extensive empirical analysis on scalability and accuracy.

Contribution

It presents a novel inconsistency measure suitable for streaming data and develops approximations for it and other measures, enabling large-scale inconsistency assessment.

Findings

Large-scale inconsistency measurement is feasible for certain measures.

The new inconsistency measure and its approximation perform well in scalability.

Empirical analysis shows trade-offs between runtime and accuracy.

Abstract

We investigate the problem of inconsistency measurement on large knowledge bases by considering stream-based inconsistency measurement, i.e., we investigate inconsistency measures that cannot consider a knowledge base as a whole but process it within a stream. For that, we present, first, a novel inconsistency measure that is apt to be applied to the streaming case and, second, stream-based approximations for the new and some existing inconsistency measures. We conduct an extensive empirical analysis on the behavior of these inconsistency measures on large knowledge bases, in terms of runtime, accuracy, and scalability. We conclude that for two of these measures, the approximation of the new inconsistency measure and an approximation of the contension inconsistency measure, large-scale inconsistency measurement is feasible.