This work describes experiments which probe the hidden representations of several BERT-style models for morphological content. The goal is to examine the extent to which discrete linguistic structure, in the form of morphological features and feature values, presents itself in the vector representations and attention distributions of pre-trained language models for five European languages. The experiments contained herein show that (i) Transformer architectures largely partition their embedding space into convex sub-regions highly correlated with morphological feature value, (ii) the contextualized nature of transformer embeddings allows models to distinguish ambiguous morphological forms in many, but not all cases, and (iii) very specific attention head/layer combinations appear to hone in on subject-verb agreement.
University of Ljubljana, Faculty of Computer and Information Science（卢布尔雅那大学计算机与信息科学学院）
Deep learning approaches are superior in NLP due to their ability to extract informative features and patterns from languages. The two most successful neural architectures are LSTM and transformers, used in large pretrained language models such as BERT. While cross-lingual approaches are on the rise, most current NLP techniques are designed and applied to English, and less-resourced languages are lagging behind. In morphologically rich languages, information is conveyed through morphology, e.g., through affixes modifying stems of words. Existing neural approaches do not explicitly use the info…
Institute of Computer Science, University of Tartu（塔尔图大学计算机科学研究所）
Recently, large pre-trained language models, such as BERT, have reached state-of-the-art performance in many natural language processing tasks, but for many languages, including Estonian, BERT models are not yet available. However, there exist several multilingual BERT models that can handle multiple languages simultaneously and that have been trained also on Estonian data. In this paper, we evaluate four multilingual models -- multilingual BERT, multilingual distilled BERT, XLM and XLM-RoBERTa -- on several NLP tasks including POS and morphological tagging, NER and text classification. Our ai…
DICTA: Israel Center for Text Analysis（DICTA：以色列文本分析中心）；Bar-Ilan University（巴伊兰大学）
We present a new pre-trained language model (PLM) for Rabbinic Hebrew, termed Berel (BERT Embeddings for Rabbinic-Encoded Language). Whilst other PLMs exist for processing Hebrew texts (e.g., HeBERT, AlephBert), they are all trained on modern Hebrew texts, which diverges substantially from Rabbinic Hebrew in terms of its lexicographical, morphological, syntactic and orthographic norms. We demonstrate the superiority of Berel on Rabbinic texts via a challenge set of Hebrew homographs. We release the new model and homograph challenge set for unrestricted use.
Can pretrained language models (PLMs) generate derivationally complex words? We present the first study investigating this question, taking BERT as the example PLM. We examine BERT's derivational capabilities in different settings, ranging from using the unmodified pretrained model to full finetuning. Our best model, DagoBERT (Derivationally and generatively optimized BERT), clearly outperforms the previous state of the art in derivation generation (DG). Furthermore, our experiments show that the input segmentation crucially impacts BERT's derivational knowledge, suggesting that the performanc…
How does the input segmentation of pretrained language models (PLMs) affect their interpretations of complex words? We present the first study investigating this question, taking BERT as the example PLM and focusing on its semantic representations of English derivatives. We show that PLMs can be interpreted as serial dual-route models, i.e., the meanings of complex words are either stored or else need to be computed from the subwords, which implies that maximally meaningful input tokens should allow for the best generalization on new words. This hypothesis is confirmed by a series of semantic …
Smart Persian Medical Assistant Group（智能波斯医疗助理小组）；Sharif University of Technology（谢里夫理工大学）
We have released Sina-BERT, a language model pre-trained on BERT (Devlin et al., 2018) to address the lack of a high-quality Persian language model in the medical domain. SINA-BERT utilizes pre-training on a large-scale corpus of medical contents including formal and informal texts collected from a variety of online resources in order to improve the performance on health-care related tasks. We employ SINA-BERT to complete following representative tasks: categorization of medical questions, medical sentiment analysis, and medical question retrieval. For each task, we have developed Persian anno…
The content on the web is in a constant state of flux. New entities, issues, and ideas continuously emerge, while the semantics of the existing conversation topics gradually shift. In recent years, pre-trained language models like BERT greatly improved the state-of-the-art for a large spectrum of content understanding tasks. Therefore, in this paper, we aim to study how these language models can be adapted to better handle continuously evolving web content. In our study, we first analyze the evolution of 2013 - 2019 Twitter data, and unequivocally confirm that a BERT model trained on past twee…
Indraprastha Institute of Information Technology, Delhi（德里英德拉普拉斯塔信息技术学院）；Extramarks Education pvt. ltd.（Extramarks教育私人有限公司）
Online education platforms are powered by various NLP pipelines, which utilize models like BERT to aid in content curation. Since the inception of the pre-trained language models like BERT, there have also been many efforts toward adapting these pre-trained models to specific domains. However, there has not been a model specifically adapted for the education domain (particularly K-12) across subjects to the best of our knowledge. In this work, we propose to train a language model on a corpus of data curated by us across multiple subjects from various sources for K-12 education. We also evaluat…
Although large pre-trained language models have achieved great success in many NLP tasks, it has been shown that they reflect human biases from their pre-training corpora. This bias may lead to undesirable outcomes when these models are applied in real-world settings. In this paper, we investigate the bias present in monolingual BERT models across a diverse set of languages (English, Greek, and Persian). While recent research has mostly focused on gender-related biases, we analyze religious and ethnic biases as well and propose a template-based method to measure any kind of bias, based on sent…
Center for Artificial Intelligence Research (CAiRE)（人工智能研究中心）；The Hong Kong University of Science and Technology（香港科技大学）
Nowadays, offensive content in social media has become a serious problem, and automatically detecting offensive language is an essential task. In this paper, we build an offensive language detection system, which combines multi-task learning with BERT-based models. Using a pre-trained language model such as BERT, we can effectively learn the representations for noisy text in social media. Besides, to boost the performance of offensive language detection, we leverage the supervision signals from other related tasks. In the OffensEval-2020 competition, our model achieves 91.51% F1 score in Engli…
International Institute of Information Technology - Hyderabad（海得拉巴国际信息技术学院）
Hateful and Toxic content has become a significant concern in today's world due to an exponential rise in social media. The increase in hate speech and harmful content motivated researchers to dedicate substantial efforts to the challenging direction of hateful content identification. In this task, we propose an approach to automatically classify hate speech and offensive content. We have used the datasets obtained from FIRE 2019 and 2020 shared tasks. We perform experiments by taking advantage of transfer learning models. We observed that the pre-trained BERT model and the multilingual-BERT m…
Never Stop Research；China Electronic Technology Group Corporation Information Science Academy（中国电子科技集团公司信息科学研究院）
We present a simple methods to leverage the table content for the BERT-based model to solve the text-to-SQL problem. Based on the observation that some of the table content match some words in question string and some of the table header also match some words in question string, we encode two addition feature vector for the deep model. Our methods also benefit the model inference in testing time as the tables are almost the same in training and testing time. We test our model on the WikiSQL dataset and outperform the BERT-based baseline by 3.7% in logic form and 3.7% in execution accuracy and …
We propose a new architecture for diacritics restoration based on contextualized embeddings, namely BERT, and we evaluate it on 12 languages with diacritics. Furthermore, we conduct a detailed error analysis on Czech, a morphologically rich language with a high level of diacritization. Notably, we manually annotate all mispredictions, showing that roughly 44% of them are actually not errors, but either plausible variants (19%), or the system corrections of erroneous data (25%). Finally, we categorize the real errors in detail. We release the code at https://github.com/ufal/bert-diacritics-rest…
Institute of Computer Science（计算机科学研究所）；University of Tartu（塔尔图大学）
This paper presents EstBERT, a large pretrained transformer-based language-specific BERT model for Estonian. Recent work has evaluated multilingual BERT models on Estonian tasks and found them to outperform the baselines. Still, based on existing studies on other languages, a language-specific BERT model is expected to improve over the multilingual ones. We first describe the EstBERT pretraining process and then present the results of the models based on finetuned EstBERT for multiple NLP tasks, including POS and morphological tagging, named entity recognition and text classification. The eval…
We propose a practical scheme to train a single multilingual sequence labeling model that yields state of the art results and is small and fast enough to run on a single CPU. Starting from a public multilingual BERT checkpoint, our final model is 6x smaller and 27x faster, and has higher accuracy than a state-of-the-art multilingual baseline. We show that our model especially outperforms on low-resource languages, and works on codemixed input text without being explicitly trained on codemixed examples. We showcase the effectiveness of our method by reporting on part-of-speech tagging and morph…
Morphosyntactic probing of multilingual BERT models
ELKH Institute for Computer Science and Control (SZTAKI)（ELKH计算机科学与控制研究所（SZTAKI））；Budapest University of Technology and Economics（布达佩斯技术与经济大学）；Comenius University in Bratislava（布拉迪斯拉发考门斯基大学）；Hebrew University of Jerusalem（耶路撒冷希伯来大学）；University of Washington（华盛顿大学）；Allen Institute for Artificial Intelligence（艾伦人工智能研究所）
Since the appearance of BERT, recent works including XLNet and RoBERTa utilize sentence embedding models pre-trained by large corpora and a large number of parameters. Because such models have large hardware and a huge amount of data, they take a long time to pre-train. Therefore it is important to attempt to make smaller models that perform comparatively. In this paper, we trained a Korean-specific model KR-BERT, utilizing a smaller vocabulary and dataset. Since Korean is one of the morphologically rich languages with poor resources using non-Latin alphabets, it is also important to capture l…
Center for Information and Language Processing, LMU Munich（慕尼黑大学信息与语言处理中心）；Faculty of Mathematics and Physics, Charles University（查理大学数学与物理学院）
Multilingual BERT (mBERT) provides sentence representations for 104 languages, which are useful for many multi-lingual tasks. Previous work probed the cross-linguality of mBERT using zero-shot transfer learning on morphological and syntactic tasks. We instead focus on the semantic properties of mBERT. We show that mBERT representations can be split into a language-specific component and a language-neutral component, and that the language-neutral component is sufficiently general in terms of modeling semantics to allow high-accuracy word-alignment and sentence retrieval but is not yet good enou…
In this work, we release COVID-Twitter-BERT (CT-BERT), a transformer-based model, pretrained on a large corpus of Twitter messages on the topic of COVID-19. Our model shows a 10-30% marginal improvement compared to its base model, BERT-Large, on five different classification datasets. The largest improvements are on the target domain. Pretrained transformer models, such as CT-BERT, are trained on a specific target domain and can be used for a wide variety of natural language processing tasks, including classification, question-answering and chatbots. CT-BERT is optimised to be used on COVID-19…
Indian Institute of Information Technology, Kalyani（印度信息技术学院卡利亚尼分校）
Social media has seen a worrying rise in hate speech in recent times. Branching to several distinct categories of cyberbullying, gender discrimination, or racism, the combined label for such derogatory content can be classified as toxic content in general. This paper presents experimentation with a Keras wrapped lightweight BERT model to successfully identify hate speech and predict probabilistic impact score for the same to extract the hateful words within sentences. The dataset used for this task is the Hate Speech and Offensive Content Detection (HASOC 2021) data from FIRE 2021 in English. …
Computer Science Department, Bar Ilan University（巴伊兰大学计算机科学系）；Allen Institute for Artificial Intelligence（艾伦人工智能研究所）
I assess the extent to which the recently introduced BERT model captures English syntactic phenomena, using (1) naturally-occurring subject-verb agreement stimuli; (2) "coloreless green ideas" subject-verb agreement stimuli, in which content words in natural sentences are randomly replaced with words sharing the same part-of-speech and inflection; and (3) manually crafted stimuli for subject-verb agreement and reflexive anaphora phenomena. The BERT model performs remarkably well on all cases.
Department of Computer Science；Engineering, Indian Institute of Technology, Kharagpur, West Bengal, India
Online hatred is a growing concern on many social media platforms. To address this issue, different social media platforms have introduced moderation policies for such content. They also employ moderators who can check the posts violating moderation policies and take appropriate action. Academicians in the abusive language research domain also perform various studies to detect such content better. Although there is extensive research in abusive language detection in English, there is a lacuna in abusive language detection in low resource languages like Hindi, Urdu etc. In this FIRE 2021 shared…
David R. Cheriton School of Computer Science（大卫·R·切里顿计算机科学学院）；University of Waterloo（滑铁卢大学）
We present, to our knowledge, the first application of BERT to document classification. A few characteristics of the task might lead one to think that BERT is not the most appropriate model: syntactic structures matter less for content categories, documents can often be longer than typical BERT input, and documents often have multiple labels. Nevertheless, we show that a straightforward classification model using BERT is able to achieve the state of the art across four popular datasets. To address the computational expense associated with BERT inference, we distill knowledge from BERT-large to…
Penn State University（宾夕法尼亚州立大学）；Worcester Polytechnic Institute（伍斯特理工学院）；University of Arkansas（阿肯色大学）；K12.com
Educational content labeled with proper knowledge components (KCs) are particularly useful to teachers or content organizers. However, manually labeling educational content is labor intensive and error-prone. To address this challenge, prior research proposed machine learning based solutions to auto-label educational content with limited success. In this work, we significantly improve prior research by (1) expanding the input types to include KC descriptions, instructional video titles, and problem descriptions (i.e., three types of prediction task), (2) doubling the granularity of the predict…
A number of morphology-based word embedding models were introduced in recent years. However, their evaluation was mostly limited to English, which is known to be a morphologically simple language. In this paper, we explore whether and to what extent incorporating morphology into word embeddings improves performance on downstream NLP tasks, in the case of morphologically rich Russian language. NLP tasks of our choice are POS tagging, Chunking, and NER -- for Russian language, all can be mostly solved using only morphology without understanding the semantics of words. Our experiments show that m…
We introduce a multilabel probing task to assess the morphosyntactic representations of word embeddings from multilingual language models. We demonstrate this task with multilingual BERT (Devlin et al., 2018), training probes for seven typologically diverse languages of varying morphological complexity: Afrikaans, Croatian, Finnish, Hebrew, Korean, Spanish, and Turkish. Through this simple but robust paradigm, we show that multilingual BERT renders many morphosyntactic features easily and simultaneously extractable (e.g., gender, grammatical case, pronominal type). We further evaluate the prob…
Linguistic Code-switching (CS) is still an understudied phenomenon in natural language processing. The NLP community has mostly focused on monolingual and multi-lingual scenarios, but little attention has been given to CS in particular. This is partly because of the lack of resources and annotated data, despite its increasing occurrence in social media platforms. In this paper, we aim at adapting monolingual models to code-switched text in various tasks. Specifically, we transfer English knowledge from a pre-trained ELMo model to different code-switched language pairs (i.e., Nepali-English, Sp…
Tel Aviv University（特拉维夫大学）；Bar Ilan University（巴伊兰大学）
Large pretrained language models (PLMs) typically tokenize the input string into contiguous subwords before any pretraining or inference. However, previous studies have claimed that this form of subword tokenization is inadequate for processing morphologically-rich languages (MRLs). We revisit this hypothesis by pretraining a BERT-style masked language model over character sequences instead of word-pieces. We compare the resulting model, dubbed TavBERT, against contemporary PLMs based on subwords for three highly complex and ambiguous MRLs (Hebrew, Turkish, and Arabic), testing them on both mo…
Salesforce Research（Salesforce研究院）；The Chinese University of Hong Kong（香港中文大学）
Visual dialog is a challenging vision-language task, where a dialog agent needs to answer a series of questions through reasoning on the image content and dialog history. Prior work has mostly focused on various attention mechanisms to model such intricate interactions. By contrast, in this work, we propose VD-BERT, a simple yet effective framework of unified vision-dialog Transformer that leverages the pretrained BERT language models for Visual Dialog tasks. The model is unified in that (1) it captures all the interactions between the image and the multi-turn dialog using a single-stream Tran…
This paper presents our submission for the SMM4H 2022-Shared Task on the classification of self-reported intimate partner violence on Twitter (in English). The goal of this task was to accurately determine if the contents of a given tweet demonstrated someone reporting their own experience with intimate partner violence. The submitted system is an ensemble of five RoBERTa models each weighted by their respective F1-scores on the validation data-set. This system performed 13% better than the baseline and was the best performing system overall for this shared task.
The Arabic language is a morphologically rich language with relatively few resources and a less explored syntax compared to English. Given these limitations, Arabic Natural Language Processing (NLP) tasks like Sentiment Analysis (SA), Named Entity Recognition (NER), and Question Answering (QA), have proven to be very challenging to tackle. Recently, with the surge of transformers based models, language-specific BERT based models have proven to be very efficient at language understanding, provided they are pre-trained on a very large corpus. Such models were able to set new standards and achiev…
Bridging the gap in online hate speech detection: a comparative analysis of BERT and traditional models for homophobic content identification on X/Twitter
Semitic languages can be highly ambiguous, having several interpretations of the same surface forms, and morphologically rich, having many morphemes that realize several morphological features. This is further exacerbated for dialectal content, which is more prone to noise and lacks a standard orthography. The morphological features can be lexicalized, like lemmas and diacritized forms, or non-lexicalized, like gender, number, and part-of-speech tags, among others. Joint modeling of the lexicalized and non-lexicalized features can identify more intricate morphological patterns, which provide b…
MahaParaphrase: A Marathi Paraphrase Detection Corpus and BERT-based Models
Pune Institute of Computer Technology（普那计算机技术研究所）；MKSSS’ Cummins College of Engineering for Women（MKSSS女子工程学院）；Indian Institute of Technology Madras（印度理工学院马德拉斯分校）；L3Cube Labs（L3Cube实验室）
Abstract: In this paper we present an approach to develop a text-classification model which would be able to identify populist content in text. The developed BERT-based model is largely successful in identifying populist content in text and produces only a negligible amount of False Negatives, which makes it well-suited as a content analysis automation tool, which shortlists potentially relevant content for human validation.
Tokenization is an important text preprocessing step to prepare input tokens for deep language models. WordPiece and BPE are de facto methods employed by important models, such as BERT and GPT. However, the impact of tokenization can be different for morphologically rich languages, such as Turkic languages, where many words can be generated by adding prefixes and suffixes. We compare five tokenizers at different granularity levels, i.e. their outputs vary from smallest pieces of characters to the surface form of words, including a Morphological-level tokenizer. We train these tokenizers and pr…
The problem of online offensive language limits the health and security of online users. It is essential to apply the latest state-of-the-art techniques in developing a system to detect online offensive language and to ensure social justice to the online communities. Our study investigates the effects of fine-tuning across several Arabic offensive language datasets. We develop multiple classifiers that use four datasets individually and in combination in order to gain knowledge about online Arabic offensive content and classify users comments accordingly. Our results demonstrate the limited ef…
Summarization of speech is a difficult problem due to the spontaneity of the flow, disfluencies, and other issues that are not usually encountered in written texts. Our work presents the first application of the BERTSum model to conversational language. We generate abstractive summaries of narrated instructional videos across a wide variety of topics, from gardening and cooking to software configuration and sports. In order to enrich the vocabulary, we use transfer learning and pretrain the model on a few large cross-domain datasets in both written and spoken English. We also do preprocessing …
For a news content distribution platform like Dailyhunt, Named Entity Recognition is a pivotal task for building better user recommendation and notification algorithms. Apart from identifying names, locations, organisations from the news for 13+ Indian languages and use them in algorithms, we also need to identify n-grams which do not necessarily fit in the definition of Named-Entity, yet they are important. For example, "me too movement", "beef ban", "alwar mob lynching". In this exercise, given an English language text, we are trying to detect case-less n-grams which convey important informa…
Offensive Language detection in social media platforms has been an active field of research over the past years. In non-native English spoken countries, social media users mostly use a code-mixed form of text in their posts/comments. This poses several challenges in the offensive content identification tasks, and considering the low resources available for Tamil, the task becomes much harder. The current study presents extensive experiments using multiple deep learning, and transfer learning models to detect offensive content on YouTube. We propose a novel and flexible approach of selective tr…
In the last two decades, automatic extractive text summarization on lectures has demonstrated to be a useful tool for collecting key phrases and sentences that best represent the content. However, many current approaches utilize dated approaches, producing sub-par outputs or requiring several hours of manual tuning to produce meaningful results. Recently, new machine learning architectures have provided mechanisms for extractive summarization through the clustering of output embeddings from deep learning models. This paper reports on the project called Lecture Summarization Service, a python b…
CNRS（法国国家科学研究中心）；Institut Polytechnique de Paris（巴黎综合理工学院）
Generated hateful and toxic content by a portion of users in social media is a rising phenomenon that motivated researchers to dedicate substantial efforts to the challenging direction of hateful content identification. We not only need an efficient automatic hate speech detection model based on advanced machine learning and natural language processing, but also a sufficiently large amount of annotated data to train a model. The lack of a sufficient amount of labelled hate speech data, along with the existing biases, has been the main issue in this domain of research. To address these needs, i…
