DeepRank: A New Deep Architecture for Relevance Ranking in Information Retrieval

  title={DeepRank: A New Deep Architecture for Relevance Ranking in Information Retrieval},
  author={Liang Pang and Yanyan Lan and J. Guo and Jun Xu and Jingfang Xu and Xueqi Cheng},
  journal={Proceedings of the 2017 ACM on Conference on Information and Knowledge Management},
  • Liang Pang, Yanyan Lan, Xueqi Cheng
  • Published 16 October 2017
  • Computer Science
  • Proceedings of the 2017 ACM on Conference on Information and Knowledge Management
This paper concerns a deep learning approach to relevance ranking in information retrieval (IR. [] Key Method Firstly, a detection strategy is designed to extract the relevant contexts. Then, a measure network is applied to determine the local relevances by utilizing a convolutional neural network (CNN) or two-dimensional gated recurrent units (2D-GRU). Finally, an aggregation network with sequential integration and term gating mechanism is used to produce a global relevance score. DeepRank well captures…

Figures and Tables from this paper

A New Deep Neural Network Based Learning to Rank Method for Information Retrieval
This work proposes a new deep neural network based method directly using extracted IR features, which is more robust and correlated to the final ranking task, and takes both local and global perspectives on documents and query into account to produce an accurate ranking.
Investigating Weak Supervision in Deep Ranking
A cascade ranking framework is proposed to combine the two weakly supervised relevance, which significantly promotes the ranking performance of neural ranking models and outperforms the best result in the last NTCIR-13 The authors Want Web (WWW) task.
DeepTileBars: Visualizing Term Distribution for Neural Information Retrieval
Although its design and implementation are light-weight, DeepTileBars outperforms other state-of-the-art Neu-IR models on benchmark datasets including the Text REtrieval Conference (TREC) 2010-2012 Web Tracks and LETOR 4.0.
A Deep Relevance Model for Zero-Shot Document Filtering
Experiments on two document collections of two different tasks demonstrate that DAZER significantly outperforms the existing alternative solutions, including the state-of-the-art deep relevance ranking models.
An Attention-based Deep Relevance Model for Few-shot Document Filtering
This article proposes a novel attention-based deep relevance model for few-shot document filtering (named ADRM), inspired by the relevance feedback methodology proposed for ad hoc retrieval, and shows that it consistently outperforms the existing technical alternatives.
Neural Ranking Models for Document Retrieval
This paper compares the proposed models in the literature along different dimensions in order to understand the major contributions and limitations of each model, analyzes the promising neural components, and proposes future research directions.
Integrated Learning of Features and Ranking Function in Information Retrieval
This paper proposes an integrated end-to-end learning framework based on learning- to-rank (L2R) to learn both neural features and the L2R ranking function simultaneously, and has the flexibility to integrate arbitrary traditional features.
Learning Groupwise Scoring Functions Using Deep Neural Networks
While in a classification or a regression setting a label or a value is assigned to each individual document, in a ranking setting we determine the relevance ordering of the entire input document


A Deep Relevance Matching Model for Ad-hoc Retrieval
A novel deep relevance matching model (DRMM) for ad-hoc retrieval that employs a joint deep architecture at the query term level for relevance matching and can significantly outperform some well-known retrieval models as well as state-of-the-art deep matching models.
Learning deep structured semantic models for web search using clickthrough data
A series of new latent semantic models with a deep structure that project queries and documents into a common low-dimensional space where the relevance of a document given a query is readily computed as the distance between them are developed.
Learning to Rank Short Text Pairs with Convolutional Deep Neural Networks
This paper presents a convolutional neural network architecture for reranking pairs of short texts, where the optimal representation of text pairs and a similarity function to relate them in a supervised way from the available training data are learned.
Match-SRNN: Modeling the Recursive Matching Structure with Spatial RNN
It is shown that, after degenerated to the exact matching scenario, Match-SRNN can approximate the dynamic programming process of longest common subsequence, and there exists a clear interpretation for Match- SRNN.
Top-k learning to rank: labeling, ranking and evaluation
A novel top-k labeling strategy which adopts the pairwise preference judgment to generate the top k ordering items from n documents in a manner similar to that of HeapSort, and the complexity of judgment is reduced to O(n log k).
Learning to rank: from pairwise approach to listwise approach
It is proposed that learning to rank should adopt the listwise approach in which lists of objects are used as 'instances' in learning, and introduces two probability models, respectively referred to as permutation probability and top k probability, to define a listwise loss function for learning.
Learning to rank for information retrieval
Three major approaches to learning to rank are introduced, i.e., the pointwise, pairwise, and listwise approaches, the relationship between the loss functions used in these approaches and the widely-used IR evaluation measures are analyzed, and the performance of these approaches on the LETOR benchmark datasets is evaluated.
Optimizing search engines using clickthrough data
The goal of this paper is to develop a method that utilizes clickthrough data for training, namely the query-log of the search engine in connection with the log of links the users clicked on in the presented ranking.
A Study of MatchPyramid Models on Ad-hoc Retrieval
The MatchPyramid models can significantly outperform several recently introduced deep matching models on the retrieval task, but still cannot compete with the traditional retrieval models, such as BM25 and language models.
Positional language models for information retrieval
A novel positional language model (PLM) is proposed which implements both heuristics in a unified language model and is effective for passage retrieval and performs better than a state-of-the-art proximity-based retrieval model.