# Deep Neural Network for Learning to Rank Query-Text Pairs

## TL;DR

This paper introduces ConvRankNet, a deep learning model combining Siamese CNN and RankNet for document ranking, demonstrating improved performance and efficient training in information retrieval tasks.

## Contribution

The paper presents ConvRankNet, a novel end-to-end deep neural network for Learning to Rank that outperforms existing feature-based models and justifies its linear test-time complexity.

## Key findings

- ConvRankNet outperforms existing feature-based models on OHSUMED.
- The model achieves efficient training and inference with linear complexity.
- Experiments validate the effectiveness of combining Siamese CNN with RankNet.

## Abstract

This paper considers the problem of document ranking in information retrieval systems by Learning to Rank. We propose ConvRankNet combining a Siamese Convolutional Neural Network encoder and the RankNet ranking model which could be trained in an end-to-end fashion. We prove a general result justifying the linear test-time complexity of pairwise Learning to Rank approach. Experiments on the OHSUMED dataset show that ConvRankNet outperforms systematically existing feature-based models.