10,548 research outputs found
Probing BERT for Ranking Abilities
No full textContextual models like BERT are highly effective in numerous text-ranking tasks. However, it is still unclear as to whether contextual models understand well-established notions of relevance that are central to IR. In this paper, we use probing, a recent approach used to analyze language models, to investigate the ranking abilities of BERT-based rankers. Most of the probing literature has focussed on linguistic and knowledge-aware capabilities of models or axiomatic analysis of ranking models. In this paper, we fill an important gap in the information retrieval literature by conducting a layer-wise probing analysis using four probes based on lexical matching, semantic similarity as well as linguistic properties like coreference resolution and named entity recognition. Our experiments show an interesting trend that BERT-rankers better encode ranking abilities at intermediate layers. Based on our observations, we train a ranking model by augmenting the ranking data with the probe data to show initial yet consistent performance improvements (The code is available at https://github.com/yolomeus/probing-search/ ).Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Web Information System
Germinação de Stevia rebaudiana Bert
No full textDissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Biologia
Querétaro, Estación del F. C. Nacional
No full textTarjeta postal ilustrada sin bordes, blanco y negro, 9 cm x 14 cm. Orientación horizontal. Reverso dividido.Vista en perspectiva de la Estación del F. C. Nacional ubicada en Querétaro, principios del siglo XX. Se observa una concurrida estación, las vías férreas y el ferrocarril.Transcripción de textos en el anverso: con texto impreso -- "Querétaro - Estación del F. C. Nacional. / Latapi y Bert Apartado 922. México." | Sin texto escritoTranscripción de textos en el reverso: sin texto impreso | Con texto escrito – "1864"Características de digitalización: equipo EPSON Scan GT-20000; resolución: 1600 ppp; formato de archivo: TIFF; profundidad de bits: 24; unidad de compresión: sin compresión; perfil de color: RGB.Características de edición: se colocó una pleca en color negro sobre la que se agregó la información del fondo documental al que pertenece así como la institución que lo resguarda. Se redimensionaron los archivos originales a 300 ppp con Adobe Photoshop CC 2018, se convirtieron a formato JPEG 2000 con compresión sin pérdida a partir de XnView Classic for Windows
Bert Wiedemann
No full textVH-MMO Fokker Friendship F.27-200 (F.27-2013). Construction number 10146, the 6th aircraft ordered new by Ansett-ANA - June 28, 1958. Operated Ansett's inaugural Alice Springs-Tennant Creek-Katherine-Darwin service - February 5, 1981.
Leased to Airlines of Northern Australia - December 1981.
Became first aircraft to be painted in the Airlines of Northern Australia livery.
Operated Ansett's final F.27 service Alice Springs-Ayers Rock-Alice Springs - March 6, 1983.Wiedemann, Bert.Date:1981-1
BERT Rankers are Brittle: A Study using Adversarial Document Perturbations
No full textContextual ranking models based on BERT are now well established for a wide range of passage and document ranking tasks. However, the robustness of BERT-based ranking models under adversarial inputs is under-explored. In this paper, we argue that BERT-rankers are not immune to adversarial attacks targeting retrieved documents given a query. Firstly, we propose algorithms for adversarial perturbation of both highly relevant and non-relevant documents using gradient-based optimization methods. The aim of our algorithms is to add/replace a small number of tokens to a highly relevant or non-relevant document to cause a large rank demotion or promotion. Our experiments show that a small number of tokens can already result in a large change in the rank of a document. Moreover, we find that BERT-rankers heavily rely on the document start/head for relevance prediction, making the initial part of the document more susceptible to adversarial attacks. More interestingly, we find a small set of recurring adversarial words that when added to documents result in successful rank demotion/promotion of any relevant/non-relevant document respectively. Finally, our adversarial tokens also show particular topic preferences within and across datasets, exposing potential biases from BERT pre-training or downstream datasets. Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Web Information System
Analyzing How BERT Performs Entity Matching
No full textState-of-the-art Entity Matching (EM) approaches rely on transformer architectures, such as BERT, for generating highly contextualized embeddings of terms. The embeddings are then used to predict whether pairs of entity descriptions refer to the same real-world entity. BERT-based EM models demonstrated to be effective, but act as black-boxes for the users, who have limited insight into the motivations behind their decisions. In this paper, we perform a multi-facet analysis of the components of pre-trained and fine-tuned BERT architectures applied to an EM task. The main findings resulting from our extensive experimental evaluation are (1) the fine-tuning process applied to the EM task mainly modifies the last layers of the BERT components, but in a different way on tokens belonging to descriptions of matching / non-matching entities; (2) the special structure of the EM datasets, where records are pairs of entity descriptions is recognized by BERT; (3) the pair-wise semantic similarity of tokens is not a key knowledge exploited by BERT-based EM models
BERT-deep CNN: state of the art for sentiment analysis of COVID-19 tweets
No full textThe COVID-19 pandemic has led to the emergence of social media platforms as crucial channels for the dissemination of information and public opinion. Comprehending the sentiment conveyed in tweets on COVID-19 is of paramount importance for individuals involved in policymaking, crisis management, and public health administration. This study seeks to conduct a comprehensive review of the current BERT and deep CNN models utilized in sentiment analysis of COVID-19 tweets. Additionally, the study aims to propose potential future research directions for the development of a BERT model that is both lightweight and high quality. The BERT model acquires contextual representations of words and effectively captures the intricate semantics of tweets related to COVID-19, whereas the deep CNN captures the hierarchical organization of the tweet embeddings. The performance of the model is exceptional, exceeding the current sentiment analysis methods for tweets related to COVID-19. Our study involves a comprehensive analysis of vast COVID-19 tweet datasets, wherein we establish the efficacy of the BERT-deep CNN models in precisely categorizing the sentiment of COVID-19 tweets in real time. The outcomes of the research offer significant perspectives on the public's attitudes, supporting decision-makers in comprehending the general viewpoint, detecting disinformation, and guiding emergency response tactics. Additionally, this study serves to enhance the progress of sentiment analysis methodologies within the realm of public health emergencies and establishes a standard for forthcoming investigations in the sentiment analysis of social media data pertaining to COVID-19
- …
