DE-PACRR: Exploring Layers Inside the PACRR Model

TL;DR

This paper investigates the internal workings of the PACRR neural IR model by visualizing intermediate layers and analyzing weight relevance, aiming to improve interpretability of deep learning models in information retrieval.

Contribution

It provides the first detailed analysis of PACRR's internal layers, offering insights into how neural IR models process relevance signals.

Findings

Visualization of intermediate layer outputs

Correlation between weights and relevance scores

Enhanced understanding of model interpretability

Abstract

Recent neural IR models have demonstrated deep learning's utility in ad-hoc information retrieval. However, deep models have a reputation for being black boxes, and the roles of a neural IR model's components may not be obvious at first glance. In this work, we attempt to shed light on the inner workings of a recently proposed neural IR model, namely the PACRR model, by visualizing the output of intermediate layers and by investigating the relationship between intermediate weights and the ultimate relevance score produced. We highlight several insights, hoping that such insights will be generally applicable.