Most previous work on the recently developed language-modeling approach to information retrieval focuses on document-specific characteristics, and therefore does not take into account the structure of the surrounding corpus. We propose a novel algorithmic framework in which information provided by document-based language models is enhanced by the incorporation of information drawn from clusters of similar documents. Using this framework, we develop a suite of new algorithms. Even the simplest typically outperforms the standard language-modeling approach in precision and recall, and our new interpolation algorithm posts statistically significant improvements for both metrics over all three corpora tested.
Evaluating Generative Ad Hoc Information Retrieval
Leipzig University（莱比锡大学）；The University of Queensland（昆士兰大学）；Bauhaus-Universität Weimar（魏玛包豪斯大学）；Friedrich-Schiller-Universität Jena（弗里德里希·席勒大学耶拿分校）；University of Kassel（卡塞尔大学）
David R. Cheriton School of Computer Science（戴维·R·切里顿计算机科学学院）；University of Waterloo（滑铁卢大学）
Following recent successes in applying BERT to question answering, we explore simple applications to ad hoc document retrieval. This required confronting the challenge posed by documents that are typically longer than the length of input BERT was designed to handle. We address this issue by applying inference on sentences individually, and then aggregating sentence scores to produce document scores. Experiments on TREC microblog and newswire test collections show that our approach is simple yet effective, as we report the highest average precision on these datasets by neural approaches that we…
Language Technologies Institute（语言技术研究所）；Carnegie Mellon University（卡内基梅隆大学）
Recent research demonstrates the effectiveness of using fine-tuned language models~(LM) for dense retrieval. However, dense retrievers are hard to train, typically requiring heavily engineered fine-tuning pipelines to realize their full potential. In this paper, we identify and address two underlying problems of dense retrievers: i)~fragility to training data noise and ii)~requiring large batches to robustly learn the embedding space. We use the recently proposed Condenser pre-training architecture, which learns to condense information into the dense vector through LM pre-training. On top of i…
Recently, Nogueira et al. [2019] proposed a new approach to document expansion based on a neural Seq2Seq model, showing significant improvement on short text retrieval task. However, this approach needs a large amount of in-domain training data. In this paper, we show that this neural document expansion approach can be effectively adapted to standard IR tasks, where labels are scarce and many long documents are present.
University of Chinese Academy of Sciences（中国科学院大学）；Institute of Software, Chinese Academy of Sciences（中国科学院软件研究所）；Computer Network Information Center, Chinese Academy of Sciences（中国科学院计算机网络信息中心）；SAP SE（SAP公司）；Max Planck Institute for Informatics（马克斯·普朗克信息学研究所）
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 strong results for ad-hoc 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 that can be used with existing neural IR models …
Texts exhibit considerable stylistic variation. This paper reports an experiment where a corpus of documents (N= 75 000) is analyzed using various simple stylistic metrics. A subset (n = 1000) of the corpus has been previously assessed to be relevant for answering given information retrieval queries. The experiment shows that this subset differs significantly from the rest of the corpus in terms of the stylistic metrics studied.
Toyota Technological Institute at Chicago（芝加哥丰田工业大学）
Here we experiment with the use of information retrieval as an augmentation for pre-trained language models. The text corpus used in information retrieval can be viewed as form of episodic memory which grows over time. By augmenting GPT 2.0 with information retrieval we achieve a zero shot 15% relative reduction in perplexity on Gigaword corpus without any re-training. We also validate our IR augmentation on an event co-reference task.
Corpus-Guided Dual-Path Propagation for Graph Retrieval-Augmented Generation
College of Software Engineering, Southeast University（东南大学软件工程学院）；School of Computer Science and Engineering, Southeast University（东南大学计算机科学与工程学院）；Key Laboratory of Computer Network and Information Integration (Southeast University) Ministry of Education（东南大学计算机网络与信息集成教育部重点实验室）
University of California, Santa Barbara（加利福尼亚大学圣巴巴拉分校）
To extract answers from a large corpus, open-domain question answering (QA) systems usually rely on information retrieval (IR) techniques to narrow the search space. Standard inverted index methods such as TF-IDF are commonly used as thanks to their efficiency. However, their retrieval performance is limited as they simply use shallow and sparse lexical features. To break the IR bottleneck, recent studies show that stronger retrieval performance can be achieved by pretraining a effective paragraph encoder that index paragraphs into dense vectors. Once trained, the corpus can be pre-encoded int…
1SimpleWay.AI 2McGill University 3University of Toronto 4University of California, Los Angeles 5The Chinese University of Hong Kong 6University of Manitoba 7Universit\'e de Montr\'eal 8Boston University 9Mila - Quebec AI Institute 10CG Matrix Technology Limited 11Lakehead University 12McMaster University
We discuss implementation issues of MARIE-1, a mostly symbolic parser fully implemented, and MARIE-2, a more statistical parser partially implemented. They address a corpus of 100,000 picture captions. We argue that the mixed approach of MARIE-2 should be better for this corpus because its algorithms (not data) are simpler.
We consider open-domain queston answering (QA) where answers are drawn from either a corpus, a knowledge base (KB), or a combination of both of these. We focus on a setting in which a corpus is supplemented with a large but incomplete KB, and on questions that require non-trivial (e.g., ``multi-hop'') reasoning. We describe PullNet, an integrated framework for (1) learning what to retrieve (from the KB and/or corpus) and (2) reasoning with this heterogeneous information to find the best answer. PullNet uses an {iterative} process to construct a question-specific subgraph that contains informat…
BM25 Wins at Scale: A Scaling Study of Retrieval-Augmented Generation Paradigms
University of Science and Technology of China（中国科学技术大学）；Metastone Technology（Metastone科技）；Information Technology Research Center, Beijing Academy of Agriculture and Forestry Sciences（北京农林科学院信息技术研究中心）
CRAFT Your Dataset: Task-Specific Synthetic Dataset Generation Through Corpus Retrieval and Augmentation
Department of Computer Science, University of Copenhagen（哥本哈根大学计算机科学系）；Center for Information and Language Processing (CIS), LMU Munich（慕尼黑大学信息与语言处理中心）；Munich Center for Machine Learning (MCML)（慕尼黑机器学习中心）
We introduce and study the task of clickbait spoiling: generating a short text that satisfies the curiosity induced by a clickbait post. Clickbait links to a web page and advertises its contents by arousing curiosity instead of providing an informative summary. Our contributions are approaches to classify the type of spoiler needed (i.e., a phrase or a passage), and to generate appropriate spoilers. A large-scale evaluation and error analysis on a new corpus of 5,000 manually spoiled clickbait posts -- the Webis Clickbait Spoiling Corpus 2022 -- shows that our spoiler type classifier achieves …
Johns Hopkins University（约翰斯·霍普金斯大学）；University of Maryland（马里兰大学）；Emory University（埃默里大学）；University of Pennsylvania（宾夕法尼亚大学）
Pretrained language models have improved effectiveness on numerous tasks, including ad-hoc retrieval. Recent work has shown that continuing to pretrain a language model with auxiliary objectives before fine-tuning on the retrieval task can further improve retrieval effectiveness. Unlike monolingual retrieval, designing an appropriate auxiliary task for cross-language mappings is challenging. To address this challenge, we use comparable Wikipedia articles in different languages to further pretrain off-the-shelf multilingual pretrained models before fine-tuning on the retrieval task. We show tha…
University of Chinese Academy of Sciences（中国科学院大学）；Tsinghua University（清华大学）
Knowledge-intensive language tasks (KILT) usually require a large body of information to provide correct answers. A popular paradigm to solve this problem is to combine a search system with a machine reader, where the former retrieves supporting evidences and the latter examines them to produce answers. Recently, the reader component has witnessed significant advances with the help of large-scale pre-trained generative models. Meanwhile most existing solutions in the search component rely on the traditional ``index-retrieve-then-rank'' pipeline, which suffers from large memory footprint and di…
The popular approaches to recommendation and ad-hoc retrieval tasks are largely distinct in the literature. In this work, we argue that many recommendation problems can also be cast as ad-hoc retrieval tasks. To demonstrate this, we build a solution for the RecSys 2018 Spotify challenge by combining standard ad-hoc retrieval models and using popular retrieval tools sets. We draw a parallel between the playlist continuation task and the task of finding good expansion terms for queries in ad-hoc retrieval, and show that standard pseudo-relevance feedback can be effective as a collaborative filte…
Beijing University of Posts and Telecommunications（北京邮电大学）；Microsoft Research Asia（微软亚洲研究院）；University of Science and Technology of China（中国科学技术大学）
Product quantization (PQ) is a widely used technique for ad-hoc retrieval. Recent studies propose supervised PQ, where the embedding and quantization models can be jointly trained with supervised learning. However, there is a lack of appropriate formulation of the joint training objective; thus, the improvements over previous non-supervised baselines are limited in reality. In this work, we propose the Matching-oriented Product Quantization (MoPQ), where a novel objective Multinoulli Contrastive Loss (MCL) is formulated. With the minimization of MCL, we are able to maximize the matching probab…
Deep neural networks have been successfully applied to many text matching tasks, such as paraphrase identification, question answering, and machine translation. Although ad-hoc retrieval can also be formalized as a text matching task, few deep models have been tested on it. In this paper, we study a state-of-the-art deep matching model, namely MatchPyramid, on the ad-hoc retrieval task. The MatchPyramid model employs a convolutional neural network over the interactions between query and document to produce the matching score. We conducted extensive experiments to study the impact of different …
CorpusBrain++: A Continual Generative Pre-Training Framework for Knowledge-Intensive Language Tasks
Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所）；University of Chinese Academy of Sciences（中国科学院大学）；University of Amsterdam（阿姆斯特丹大学）
Most Information Retrieval models compute the relevance score of a document for a given query by summing term weights specific to a document or a query. Heuristic approaches, like TF-IDF, or probabilistic models, like BM25, are used to specify how a term weight is computed. In this paper, we propose to leverage learning-to-rank principles to learn how to compute a term weight for a given document based on the term occurrence pattern.
This paper describes the work towards Gujarati Ad hoc Monolingual Retrieval task for widely used Information Retrieval (IR) models. We present an indexing baseline for the Gujarati Language represented by Mean Average Precision (MAP) values. Our objective is to obtain a relative picture of a better IR model for Gujarati Language. Results show that Classical IR models like Term Frequency Inverse Document Frequency (TF_IDF) performs better when compared to few recent probabilistic IR models. The experiments helped to identify the outperforming IR models for Gujarati Language.
In recent years, deep neural networks have led to exciting breakthroughs in speech recognition, computer vision, and natural language processing (NLP) tasks. However, there have been few positive results of deep models on ad-hoc retrieval tasks. This is partially due to the fact that many important characteristics of the ad-hoc retrieval task have not been well addressed in deep models yet. Typically, the ad-hoc retrieval task is formalized as a matching problem between two pieces of text in existing work using deep models, and treated equivalent to many NLP tasks such as paraphrase identifica…
Traditional statistical retrieval models often treat each document as a whole. In many cases, however, a document is relevant to a query only because a small part of it contain the targeted information. In this work, we propose a neural passage model (NPM) that uses passage-level information to improve the performance of ad-hoc retrieval. Instead of using a single window to extract passages, our model automatically learns to weight passages with different granularities in the training process. We show that the passage-based document ranking paradigm from previous studies can be directly derive…
The ad-hoc retrieval task is to rank related documents given a query and a document collection. A series of deep learning based approaches have been proposed to solve such problem and gained lots of attention. However, we argue that they are inherently based on local word sequences, ignoring the subtle long-distance document-level word relationships. To solve the problem, we explicitly model the document-level word relationship through the graph structure, capturing the subtle information via graph neural networks. In addition, due to the complexity and scale of the document collections, it is…
In this paper, we present the experimental work done on Query Expansion (QE) for retrieval tasks of Gujarati text documents. In information retrieval, it is very difficult to estimate the exact user need, query expansion adds terms to the original query, which provides more information about the user need. There are various approaches to query expansion. In our work, manual thesaurus based query expansion was performed to evaluate the performance of widely used information retrieval models for Gujarati text documents. Results show that query expansion improves the recall of text documents.
Pre-training and fine-tuning have achieved remarkable success in many downstream natural language processing (NLP) tasks. Recently, pre-training methods tailored for information retrieval (IR) have also been explored, and the latest success is the PROP method which has reached new SOTA on a variety of ad-hoc retrieval benchmarks. The basic idea of PROP is to construct the \textit{representative words prediction} (ROP) task for pre-training inspired by the query likelihood model. Despite its exciting performance, the effectiveness of PROP might be bounded by the classical unigram language model…
Bergische Universität Wuppertal（伍珀塔尔大学）；Universität Hamburg（汉堡大学）
The corpus, from which a predictive language model is trained, can be considered the experience of a semantic system. We recorded everyday reading of two participants for two months on a tablet, generating individual corpus samples of 300/500K tokens. Then we trained word2vec models from individual corpora and a 70 million-sentence newspaper corpus to obtain individual and norm-based long-term memory structure. To test whether individual corpora can make better predictions for a cognitive task of long-term memory retrieval, we generated stimulus materials consisting of 134 sentences with uncor…
Designing pre-training objectives that more closely resemble the downstream tasks for pre-trained language models can lead to better performance at the fine-tuning stage, especially in the ad-hoc retrieval area. Existing pre-training approaches tailored for IR tried to incorporate weak supervised signals, such as query-likelihood based sampling, to construct pseudo query-document pairs from the raw textual corpus. However, these signals rely heavily on the sampling method. For example, the query likelihood model may lead to much noise in the constructed pre-training data. \blfootnote{$\dagger$…
Recently pre-trained language representation models such as BERT have shown great success when fine-tuned on downstream tasks including information retrieval (IR). However, pre-training objectives tailored for ad-hoc retrieval have not been well explored. In this paper, we propose Pre-training with Representative wOrds Prediction (PROP) for ad-hoc retrieval. PROP is inspired by the classical statistical language model for IR, specifically the query likelihood model, which assumes that the query is generated as the piece of text representative of the "ideal" document. Based on this idea, we con…
