Pub Date : 2023-02-27DOI: 10.48550/arXiv.2302.14021
Gonccalo Azevedo Mendes, Bruno Martins
The analysis of emotions expressed in text has numerous applications. In contrast to categorical analysis, focused on classifying emotions according to a pre-defined set of common classes, dimensional approaches can offer a more nuanced way to distinguish between different emotions. Still, dimensional methods have been less studied in the literature. Considering a valence-arousal dimensional space, this work assesses the use of pre-trained Transformers to predict these two dimensions on a continuous scale, with input texts from multiple languages and domains. We specifically combined multiple annotated datasets from previous studies, corresponding to either emotional lexica or short text documents, and evaluated models of multiple sizes and trained under different settings. Our results show that model size can have a significant impact on the quality of predictions, and that by fine-tuning a large model we can confidently predict valence and arousal in multiple languages. We make available the code, models, and supporting data.
{"title":"Quantifying Valence and Arousal in Text with Multilingual Pre-trained Transformers","authors":"Gonccalo Azevedo Mendes, Bruno Martins","doi":"10.48550/arXiv.2302.14021","DOIUrl":"https://doi.org/10.48550/arXiv.2302.14021","url":null,"abstract":"The analysis of emotions expressed in text has numerous applications. In contrast to categorical analysis, focused on classifying emotions according to a pre-defined set of common classes, dimensional approaches can offer a more nuanced way to distinguish between different emotions. Still, dimensional methods have been less studied in the literature. Considering a valence-arousal dimensional space, this work assesses the use of pre-trained Transformers to predict these two dimensions on a continuous scale, with input texts from multiple languages and domains. We specifically combined multiple annotated datasets from previous studies, corresponding to either emotional lexica or short text documents, and evaluated models of multiple sizes and trained under different settings. Our results show that model size can have a significant impact on the quality of predictions, and that by fine-tuning a large model we can confidently predict valence and arousal in multiple languages. We make available the code, models, and supporting data.","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121679763","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-02-23DOI: 10.48550/arXiv.2302.12139
Alexander Flick, Sebastian Jäger, Ivana Trajanovska, F. Biessmann
Extracting structured information from unstructured data is one of the key challenges in modern information retrieval applications, including e-commerce. Here, we demonstrate how recent advances in machine learning, combined with a recently published multilingual data set with standardized fine-grained product category information, enable robust product attribute extraction in challenging transfer learning settings. Our models can reliably predict product attributes across online shops, languages, or both. Furthermore, we show that our models can be used to match product taxonomies between online retailers.
{"title":"Automated Extraction of Fine-Grained Standardized Product Information from Unstructured Multilingual Web Data","authors":"Alexander Flick, Sebastian Jäger, Ivana Trajanovska, F. Biessmann","doi":"10.48550/arXiv.2302.12139","DOIUrl":"https://doi.org/10.48550/arXiv.2302.12139","url":null,"abstract":"Extracting structured information from unstructured data is one of the key challenges in modern information retrieval applications, including e-commerce. Here, we demonstrate how recent advances in machine learning, combined with a recently published multilingual data set with standardized fine-grained product category information, enable robust product attribute extraction in challenging transfer learning settings. Our models can reliably predict product attributes across online shops, languages, or both. Furthermore, we show that our models can be used to match product taxonomies between online retailers.","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123734819","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-02-20DOI: 10.48550/arXiv.2302.09947
G. Faggioli, Thibault Formal, S. Marchesin, S. Clinchant, N. Ferro, Benjamin Piwowarski
Evaluation in Information Retrieval relies on post-hoc empirical procedures, which are time-consuming and expensive operations. To alleviate this, Query Performance Prediction (QPP) models have been developed to estimate the performance of a system without the need for human-made relevance judgements. Such models, usually relying on lexical features from queries and corpora, have been applied to traditional sparse IR methods - with various degrees of success. With the advent of neural IR and large Pre-trained Language Models, the retrieval paradigm has significantly shifted towards more semantic signals. In this work, we study and analyze to what extent current QPP models can predict the performance of such systems. Our experiments consider seven traditional bag-of-words and seven BERT-based IR approaches, as well as nineteen state-of-the-art QPPs evaluated on two collections, Deep Learning '19 and Robust '04. Our findings show that QPPs perform statistically significantly worse on neural IR systems. In settings where semantic signals are prominent (e.g., passage retrieval), their performance on neural models drops by as much as 10% compared to bag-of-words approaches. On top of that, in lexical-oriented scenarios, QPPs fail to predict performance for neural IR systems on those queries where they differ from traditional approaches the most.
{"title":"Query Performance Prediction for Neural IR: Are We There Yet?","authors":"G. Faggioli, Thibault Formal, S. Marchesin, S. Clinchant, N. Ferro, Benjamin Piwowarski","doi":"10.48550/arXiv.2302.09947","DOIUrl":"https://doi.org/10.48550/arXiv.2302.09947","url":null,"abstract":"Evaluation in Information Retrieval relies on post-hoc empirical procedures, which are time-consuming and expensive operations. To alleviate this, Query Performance Prediction (QPP) models have been developed to estimate the performance of a system without the need for human-made relevance judgements. Such models, usually relying on lexical features from queries and corpora, have been applied to traditional sparse IR methods - with various degrees of success. With the advent of neural IR and large Pre-trained Language Models, the retrieval paradigm has significantly shifted towards more semantic signals. In this work, we study and analyze to what extent current QPP models can predict the performance of such systems. Our experiments consider seven traditional bag-of-words and seven BERT-based IR approaches, as well as nineteen state-of-the-art QPPs evaluated on two collections, Deep Learning '19 and Robust '04. Our findings show that QPPs perform statistically significantly worse on neural IR systems. In settings where semantic signals are prominent (e.g., passage retrieval), their performance on neural models drops by as much as 10% compared to bag-of-words approaches. On top of that, in lexical-oriented scenarios, QPPs fail to predict performance for neural IR systems on those queries where they differ from traditional approaches the most.","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"115 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115193501","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-02-13DOI: 10.48550/arXiv.2302.06448
Santosh T.Y.S.S, Philipp Bock, Matthias Grabmair
Segmentation and Rhetorical Role Labeling of legal judgements play a crucial role in retrieval and adjacent tasks, including case summarization, semantic search, argument mining etc. Previous approaches have formulated this task either as independent classification or sequence labeling of sentences. In this work, we reformulate the task at span level as identifying spans of multiple consecutive sentences that share the same rhetorical role label to be assigned via classification. We employ semi-Markov Conditional Random Fields (CRF) to jointly learn span segmentation and span label assignment. We further explore three data augmentation strategies to mitigate the data scarcity in the specialized domain of law where individual documents tend to be very long and annotation cost is high. Our experiments demonstrate improvement of span-level prediction metrics with a semi-Markov CRF model over a CRF baseline. This benefit is contingent on the presence of multi sentence spans in the document.
{"title":"Joint Span Segmentation and Rhetorical Role Labeling with Data Augmentation for Legal Documents","authors":"Santosh T.Y.S.S, Philipp Bock, Matthias Grabmair","doi":"10.48550/arXiv.2302.06448","DOIUrl":"https://doi.org/10.48550/arXiv.2302.06448","url":null,"abstract":"Segmentation and Rhetorical Role Labeling of legal judgements play a crucial role in retrieval and adjacent tasks, including case summarization, semantic search, argument mining etc. Previous approaches have formulated this task either as independent classification or sequence labeling of sentences. In this work, we reformulate the task at span level as identifying spans of multiple consecutive sentences that share the same rhetorical role label to be assigned via classification. We employ semi-Markov Conditional Random Fields (CRF) to jointly learn span segmentation and span label assignment. We further explore three data augmentation strategies to mitigate the data scarcity in the specialized domain of law where individual documents tend to be very long and annotation cost is high. Our experiments demonstrate improvement of span-level prediction metrics with a semi-Markov CRF model over a CRF baseline. This benefit is contingent on the presence of multi sentence spans in the document.","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"104 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115586601","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-02-12DOI: 10.48550/arXiv.2302.05990
Alejandro Ariza-Casabona, Bartlomiej Twardowski, T. Wijaya
Multi-domain recommender systems benefit from cross-domain representation learning and positive knowledge transfer. Both can be achieved by introducing a specific modeling of input data (i.e. disjoint history) or trying dedicated training regimes. At the same time, treating domains as separate input sources becomes a limitation as it does not capture the interplay that naturally exists between domains. In this work, we efficiently learn multi-domain representation of sequential users' interactions using graph neural networks. We use temporal intra- and inter-domain interactions as contextual information for our method called MAGRec (short for Multi-domAin Graph-based Recommender). To better capture all relations in a multi-domain setting, we learn two graph-based sequential representations simultaneously: domain-guided for recent user interest, and general for long-term interest. This approach helps to mitigate the negative knowledge transfer problem from multiple domains and improve overall representation. We perform experiments on publicly available datasets in different scenarios where MAGRec consistently outperforms state-of-the-art methods. Furthermore, we provide an ablation study and discuss further extensions of our method.
{"title":"Exploiting Graph Structured Cross-Domain Representation for Multi-Domain Recommendation","authors":"Alejandro Ariza-Casabona, Bartlomiej Twardowski, T. Wijaya","doi":"10.48550/arXiv.2302.05990","DOIUrl":"https://doi.org/10.48550/arXiv.2302.05990","url":null,"abstract":"Multi-domain recommender systems benefit from cross-domain representation learning and positive knowledge transfer. Both can be achieved by introducing a specific modeling of input data (i.e. disjoint history) or trying dedicated training regimes. At the same time, treating domains as separate input sources becomes a limitation as it does not capture the interplay that naturally exists between domains. In this work, we efficiently learn multi-domain representation of sequential users' interactions using graph neural networks. We use temporal intra- and inter-domain interactions as contextual information for our method called MAGRec (short for Multi-domAin Graph-based Recommender). To better capture all relations in a multi-domain setting, we learn two graph-based sequential representations simultaneously: domain-guided for recent user interest, and general for long-term interest. This approach helps to mitigate the negative knowledge transfer problem from multiple domains and improve overall representation. We perform experiments on publicly available datasets in different scenarios where MAGRec consistently outperforms state-of-the-art methods. Furthermore, we provide an ablation study and discuss further extensions of our method.","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116154329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-01-29DOI: 10.48550/arXiv.2301.12394
vStvep'an vSimsa, Milan vSulc, Maty'avs Skalick'y, Yash J. Patel, Ahmed Hamdi
The lack of data for information extraction (IE) from semi-structured business documents is a real problem for the IE community. Publications relying on large-scale datasets use only proprietary, unpublished data due to the sensitive nature of such documents. Publicly available datasets are mostly small and domain-specific. The absence of a large-scale public dataset or benchmark hinders the reproducibility and cross-evaluation of published methods. The DocILE 2023 competition, hosted as a lab at the CLEF 2023 conference and as an ICDAR 2023 competition, will run the first major benchmark for the tasks of Key Information Localization and Extraction (KILE) and Line Item Recognition (LIR) from business documents. With thousands of annotated real documents from open sources, a hundred thousand of generated synthetic documents, and nearly a million unlabeled documents, the DocILE lab comes with the largest publicly available dataset for KILE and LIR. We are looking forward to contributions from the Computer Vision, Natural Language Processing, Information Retrieval, and other communities. The data, baselines, code and up-to-date information about the lab and competition are available at https://docile.rossum.ai/.
{"title":"DocILE 2023 Teaser: Document Information Localization and Extraction","authors":"vStvep'an vSimsa, Milan vSulc, Maty'avs Skalick'y, Yash J. Patel, Ahmed Hamdi","doi":"10.48550/arXiv.2301.12394","DOIUrl":"https://doi.org/10.48550/arXiv.2301.12394","url":null,"abstract":"The lack of data for information extraction (IE) from semi-structured business documents is a real problem for the IE community. Publications relying on large-scale datasets use only proprietary, unpublished data due to the sensitive nature of such documents. Publicly available datasets are mostly small and domain-specific. The absence of a large-scale public dataset or benchmark hinders the reproducibility and cross-evaluation of published methods. The DocILE 2023 competition, hosted as a lab at the CLEF 2023 conference and as an ICDAR 2023 competition, will run the first major benchmark for the tasks of Key Information Localization and Extraction (KILE) and Line Item Recognition (LIR) from business documents. With thousands of annotated real documents from open sources, a hundred thousand of generated synthetic documents, and nearly a million unlabeled documents, the DocILE lab comes with the largest publicly available dataset for KILE and LIR. We are looking forward to contributions from the Computer Vision, Natural Language Processing, Information Retrieval, and other communities. The data, baselines, code and up-to-date information about the lab and competition are available at https://docile.rossum.ai/.","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126857642","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-01-26DOI: 10.48550/arXiv.2301.10926
Han Zhang, Ziwei Zhu, James Caverlee
Recent work in news recommendation has demonstrated that recommenders can over-expose users to articles that support their pre-existing opinions. However, most existing work focuses on a static setting or over a short-time window, leaving open questions about the long-term and dynamic impacts of news recommendations. In this paper, we explore these dynamic impacts through a systematic study of three research questions: 1) How do the news reading behaviors of users change after repeated long-term interactions with recommenders? 2) How do the inherent preferences of users change over time in such a dynamic recommender system? 3) Can the existing SOTA static method alleviate the problem in the dynamic environment? Concretely, we conduct a comprehensive data-driven study through simulation experiments of political polarization in news recommendations based on 40,000 annotated news articles. We find that users are rapidly exposed to more extreme content as the recommender evolves. We also find that a calibration-based intervention can slow down this polarization, but leaves open significant opportunities for future improvements
{"title":"Evolution of Filter Bubbles and Polarization in News Recommendation","authors":"Han Zhang, Ziwei Zhu, James Caverlee","doi":"10.48550/arXiv.2301.10926","DOIUrl":"https://doi.org/10.48550/arXiv.2301.10926","url":null,"abstract":"Recent work in news recommendation has demonstrated that recommenders can over-expose users to articles that support their pre-existing opinions. However, most existing work focuses on a static setting or over a short-time window, leaving open questions about the long-term and dynamic impacts of news recommendations. In this paper, we explore these dynamic impacts through a systematic study of three research questions: 1) How do the news reading behaviors of users change after repeated long-term interactions with recommenders? 2) How do the inherent preferences of users change over time in such a dynamic recommender system? 3) Can the existing SOTA static method alleviate the problem in the dynamic environment? Concretely, we conduct a comprehensive data-driven study through simulation experiments of political polarization in news recommendations based on 40,000 annotated news articles. We find that users are rapidly exposed to more extreme content as the recommender evolves. We also find that a calibration-based intervention can slow down this polarization, but leaves open significant opportunities for future improvements","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"170 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115419177","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-01-25DOI: 10.48550/arXiv.2301.10493
Weronika Lajewska, K. Balog
This paper reports on an effort of reproducing the organizers' baseline as well as the top performing participant submission at the 2021 edition of the TREC Conversational Assistance track. TREC systems are commonly regarded as reference points for effectiveness comparison. Yet, the papers accompanying them have less strict requirements than peer-reviewed publications, which can make reproducibility challenging. Our results indicate that key practical information is indeed missing. While the results can be reproduced within a 19% relative margin with respect to the main evaluation measure, the relative difference between the baseline and the top performing approach shrinks from the reported 18% to 5%. Additionally, we report on a new set of experiments aimed at understanding the impact of various pipeline components. We show that end-to-end system performance can indeed benefit from advanced retrieval techniques in either stage of a two-stage retrieval pipeline. We also measure the impact of the dataset used for fine-tuning the query rewriter and find that employing different query rewriting methods in different stages of the retrieval pipeline might be beneficial. Moreover, these results are shown to generalize across the 2020 and 2021 editions of the track. We conclude our study with a list of lessons learned and practical suggestions.
{"title":"From Baseline to Top Performer: A Reproducibility Study of Approaches at the TREC 2021 Conversational Assistance Track","authors":"Weronika Lajewska, K. Balog","doi":"10.48550/arXiv.2301.10493","DOIUrl":"https://doi.org/10.48550/arXiv.2301.10493","url":null,"abstract":"This paper reports on an effort of reproducing the organizers' baseline as well as the top performing participant submission at the 2021 edition of the TREC Conversational Assistance track. TREC systems are commonly regarded as reference points for effectiveness comparison. Yet, the papers accompanying them have less strict requirements than peer-reviewed publications, which can make reproducibility challenging. Our results indicate that key practical information is indeed missing. While the results can be reproduced within a 19% relative margin with respect to the main evaluation measure, the relative difference between the baseline and the top performing approach shrinks from the reported 18% to 5%. Additionally, we report on a new set of experiments aimed at understanding the impact of various pipeline components. We show that end-to-end system performance can indeed benefit from advanced retrieval techniques in either stage of a two-stage retrieval pipeline. We also measure the impact of the dataset used for fine-tuning the query rewriter and find that employing different query rewriting methods in different stages of the retrieval pipeline might be beneficial. Moreover, these results are shown to generalize across the 2020 and 2021 editions of the track. We conclude our study with a list of lessons learned and practical suggestions.","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"93 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134081561","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-01-25DOI: 10.48550/arXiv.2301.10576
Simon Lupart, S. Clinchant
Neural retrieval models have acquired significant effectiveness gains over the last few years compared to term-based methods. Nevertheless, those models may be brittle when faced to typos, distribution shifts or vulnerable to malicious attacks. For instance, several recent papers demonstrated that such variations severely impacted models performances, and then tried to train more resilient models. Usual approaches include synonyms replacements or typos injections -- as data-augmentation -- and the use of more robust tokenizers (characterBERT, BPE-dropout). To further complement the literature, we investigate in this paper adversarial training as another possible solution to this robustness issue. Our comparison includes the two main families of BERT-based neural retrievers, i.e. dense and sparse, with and without distillation techniques. We then demonstrate that one of the most simple adversarial training techniques -- the Fast Gradient Sign Method (FGSM) -- can improve first stage rankers robustness and effectiveness. In particular, FGSM increases models performances on both in-domain and out-of-domain distributions, and also on queries with typos, for multiple neural retrievers.
{"title":"A Study on FGSM Adversarial Training for Neural Retrieval","authors":"Simon Lupart, S. Clinchant","doi":"10.48550/arXiv.2301.10576","DOIUrl":"https://doi.org/10.48550/arXiv.2301.10576","url":null,"abstract":"Neural retrieval models have acquired significant effectiveness gains over the last few years compared to term-based methods. Nevertheless, those models may be brittle when faced to typos, distribution shifts or vulnerable to malicious attacks. For instance, several recent papers demonstrated that such variations severely impacted models performances, and then tried to train more resilient models. Usual approaches include synonyms replacements or typos injections -- as data-augmentation -- and the use of more robust tokenizers (characterBERT, BPE-dropout). To further complement the literature, we investigate in this paper adversarial training as another possible solution to this robustness issue. Our comparison includes the two main families of BERT-based neural retrievers, i.e. dense and sparse, with and without distillation techniques. We then demonstrate that one of the most simple adversarial training techniques -- the Fast Gradient Sign Method (FGSM) -- can improve first stage rankers robustness and effectiveness. In particular, FGSM increases models performances on both in-domain and out-of-domain distributions, and also on queries with typos, for multiple neural retrievers.","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124956429","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-01-25DOI: 10.48550/arXiv.2301.10444
Carlos Lassance, Herv'e D'ejean, S. Clinchant
Finetuning Pretrained Language Models (PLM) for IR has been de facto the standard practice since their breakthrough effectiveness few years ago. But, is this approach well understood? In this paper, we study the impact of the pretraining collection on the final IR effectiveness. In particular, we challenge the current hypothesis that PLM shall be trained on a large enough generic collection and we show that pretraining from scratch on the collection of interest is surprisingly competitive with the current approach. We benchmark first-stage ranking rankers and cross-encoders for reranking on the task of general passage retrieval on MSMARCO, Mr-Tydi for Arabic, Japanese and Russian, and TripClick for specific domain. Contrary to popular belief, we show that, for finetuning first-stage rankers, models pretrained solely on their collection have equivalent or better effectiveness compared to more general models. However, there is a slight effectiveness drop for rerankers pretrained only on the target collection. Overall, our study sheds a new light on the role of the pretraining collection and should make our community ponder on building specialized models by pretraining from scratch. Last but not least, doing so could enable better control of efficiency, data bias and replicability, which are key research questions for the IR community.
{"title":"An Experimental Study on Pretraining Transformers from Scratch for IR","authors":"Carlos Lassance, Herv'e D'ejean, S. Clinchant","doi":"10.48550/arXiv.2301.10444","DOIUrl":"https://doi.org/10.48550/arXiv.2301.10444","url":null,"abstract":"Finetuning Pretrained Language Models (PLM) for IR has been de facto the standard practice since their breakthrough effectiveness few years ago. But, is this approach well understood? In this paper, we study the impact of the pretraining collection on the final IR effectiveness. In particular, we challenge the current hypothesis that PLM shall be trained on a large enough generic collection and we show that pretraining from scratch on the collection of interest is surprisingly competitive with the current approach. We benchmark first-stage ranking rankers and cross-encoders for reranking on the task of general passage retrieval on MSMARCO, Mr-Tydi for Arabic, Japanese and Russian, and TripClick for specific domain. Contrary to popular belief, we show that, for finetuning first-stage rankers, models pretrained solely on their collection have equivalent or better effectiveness compared to more general models. However, there is a slight effectiveness drop for rerankers pretrained only on the target collection. Overall, our study sheds a new light on the role of the pretraining collection and should make our community ponder on building specialized models by pretraining from scratch. Last but not least, doing so could enable better control of efficiency, data bias and replicability, which are key research questions for the IR community.","PeriodicalId":126309,"journal":{"name":"European Conference on Information Retrieval","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128447144","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：