Detection of decision-making manipulation in the pairwise comparisons method

TL;DR

This work targets manipulation within $PC$ matrices used in $AHP$-based decision making by introducing three attack algorithms that bias rankings. It then demonstrates how convolutional neural networks, especially a 3D-CNN with a determinant-based preprocessing $D(C)$, can effectively detect manipulated matrices from synthetic data. The study reports high detection rates, including near-perfect performance for several matrix sizes, and identifies limitations of traditional inconsistency or error indicators for this task. The results emphasize the value of multi-dimensional, structure-preserving representations for manipulation detection and point to future work on richer preprocessing and more sophisticated manipulations.

Abstract

Most decision-making models, including the pairwise comparison method, assume the decision-makers honesty. However, it is easy to imagine a situation where a decision-maker tries to manipulate the ranking results. This paper presents three simple manipulation methods in the pairwise comparison method. We then try to detect these methods using appropriately constructed neural networks. Experimental results accompany the proposed solutions on the generated data, showing a considerable manipulation detection level.

Figures (5)

• Figure 1: Naive algorithm heat map
• Figure 2: Basic algorithm heat map
• Figure 3: Advanced algorithm heat map
• Figure 4: A simple (2-D) $PC$ matrix being flattened to a (1-D) vector.
• Figure 5: Design of Neural Networks used in the experiments