An end-to-end pseudo relevance feedback framework for neural document retrieval. Issue 2 (March 2020)

• Record Type: Journal Article
• Authors: Wang, Le; Luo, Ze; Li, Canjia; He, Ben; Sun, Le; Yu, Hao; Sun, Yingfei
• Is Part Of: Information processing & management. Volume 57:Issue 2(2020:Mar.)
• Publication Date: 2020-03
• Subjects: Information storage and retrieval systems -- Periodicals; Information science -- Periodicals; Systèmes d'information -- Périodiques; Sciences de l'information -- Périodiques; Information science; Information storage and retrieval systems; Periodicals; 658.4038
• DOI: 10.1016/j.ipm.2019.102182
• Languages: English
• ISSNs: 0306-4573

Abstract

Highlights: The proposal of a novel neural pseudo relevance feedback (NPRF) framework that enables the integration of PRF for neural retrieval. To the best of our knowledge, this is the first neural retrieval model that integrates PRF information by end-to-end learning. Three instantiations of the NPRF framework are introduced based on state-of-the-art neural IR models, namely the unigram DRMM and KNRM models, and the n-gram PACRR model. Thorough experiments support the framework's intuition and showcase its ability in enhancing the retrieval performance of neural retrieval. In addition, analysis indicates reduced training and validation losses by our NPRF framework, leading to significantly improved effectiveness of the learned ranking functions. Abstract: Pseudo relevance feedback (PRF) is commonly used to boost the performance of traditional information retrieval (IR) models by using top-ranked documents to identify and weight new query terms, thereby reducing the effect of query-document vocabulary mismatches. While neural retrieval models have recently demonstrated promising results for ad-hoc document retrieval, combining them with PRF is not straightforward due to incompatibilities between existing PRF approaches and neural architectures. To bridge this gap, we propose an end-to-end neural PRF framework, coined NPRF, that enriches the representation of user information need from a single query to multiple PRF documents. NPRF can be used with existing neural IR models by embedding different neural models as building blocks. Three state-of-the-art neural retrieval models, including the unigram DRMM and KNRM models, and the position-aware PACRR model, are utilized to instantiate the NPRF framework. Extensive experiments on two standard test collections, TREC1-3 and Robust04, confirm the effectiveness of the proposed NPRF framework in improving the performance of three state-of-the-art neural IR models. In addition, analysis shows that integrating the existing neural IR models within the NPRF framework results in reduced training and validation losses, and consequently, improved effectiveness of the learned ranking functions.