Evaluating Local Model-Agnostic Explanations of Learning to Rank Models with Decision Paths

TL;DR

This paper introduces a systematic method to evaluate local explanations of learning-to-rank models by comparing explanation-derived feature importance scores to ground truth scores extracted from decision paths in tree-based models.

Contribution

It proposes a novel evaluation technique for LTR explanations using decision paths in tree models, enabling direct comparison with ground truth feature importance.

Findings

Explanation accuracy varies significantly across models and data points.

Tree-based models allow extraction of ground truth feature importance scores.

Evaluation reveals inconsistencies in explanation techniques for LTR models.

Abstract

Local explanations of learning-to-rank (LTR) models are thought to extract the most important features that contribute to the ranking predicted by the LTR model for a single data point. Evaluating the accuracy of such explanations is challenging since the ground truth feature importance scores are not available for most modern LTR models. In this work, we propose a systematic evaluation technique for explanations of LTR models. Instead of using black-box models, such as neural networks, we propose to focus on tree-based LTR models, from which we can extract the ground truth feature importance scores using decision paths. Once extracted, we can directly compare the ground truth feature importance scores to the feature importance scores generated with explanation techniques. We compare two recently proposed explanation techniques for LTR models when using decision trees and gradient boosting models on the MQ2008 dataset. We show that the explanation accuracy in these techniques can largely vary depending on the explained model and even which data point is explained.