The Marathi language is one of the prominent languages used in India. It is predominantly spoken by the people of Maharashtra. Over the past decade, the usage of language on online platforms has tremendously increased. However, research on Natural Language Processing (NLP) approaches for Marathi text has not received much attention. Marathi is a morphologically rich language and uses a variant of the Devanagari script in the written form. This works aims to provide a comprehensive overview of available resources and models for Marathi text classification. We evaluate CNN, LSTM, ULMFiT, and BERT based models on two publicly available Marathi text classification datasets and present a comparative analysis. The pre-trained Marathi fast text word embeddings by Facebook and IndicNLP are used in conjunction with word-based models. We show that basic single layer models based on CNN and LSTM coupled with FastText embeddings perform on par with the BERT based models on the available datasets. We hope our paper aids focused research and experiments in the area of Marathi NLP.
Sentence semantic understanding is a key topic in the field of natural language processing. Recently, contextualized word representations derived from pre-trained language models such as ELMO and BERT have shown significant improvements for a wide range of semantic tasks, e.g. question answering, text classification and sentiment analysis. However, how to add external knowledge to further improve the semantic modeling capability of model is worth probing. In this paper, we propose a novel approach to combining syntax information with a pre-trained language model. In order to evaluate the effect of the pre-training model, first, we introduce RNN-based and Transformer-based pre-trained language models; secondly, to better integrate external knowledge, such as syntactic information integrate with the pre-training model, we propose a dependency syntax expansion (DSE) model. For evaluation, we have selected two subtasks: sentence completion task and biological relation extraction task. The experimental results show that our model achieves 91.2\% accuracy, outperforming the baseline model by 37.8\% on sentence completion task. And it also gets competitive performance by 75.1\% $F_{1}$ score on relation extraction task.
Extreme Multi-label text Classification (XMC) is a task of finding the most relevant labels from a large label set. Nowadays deep learning-based methods have shown significant success in XMC. However, the existing methods (e.g., AttentionXML and X-Transformer etc) still suffer from 1) combining several models to train and predict for one dataset, and 2) sampling negative labels statically during the process of training label ranking model, which reduces both the efficiency and accuracy of the model. To address the above problems, we proposed LightXML, which adopts end-to-end training and dynamic negative labels sampling. In LightXML, we use generative cooperative networks to recall and rank labels, in which label recalling part generates negative and positive labels, and label ranking part distinguishes positive labels from these labels. Through these networks, negative labels are sampled dynamically during label ranking part training by feeding with the same text representation. Extensive experiments show that LightXML outperforms state-of-the-art methods in five extreme multi-label datasets with much smaller model size and lower computational complexity. In particular, on the Amazon dataset with 670K labels, LightXML can reduce the model size up to 72% compared to AttentionXML.
Social media platforms like Facebook, Twitter, and Instagram have enabled connection and communication on a large scale. It has revolutionized the rate at which information is shared and enhanced its reach. However, another side of the coin dictates an alarming story. These platforms have led to an increase in the creation and spread of fake news. The fake news has not only influenced people in the wrong direction but also claimed human lives. During these critical times of the Covid19 pandemic, it is easy to mislead people and make them believe in fatal information. Therefore it is important to curb fake news at source and prevent it from spreading to a larger audience. We look at automated techniques for fake news detection from a data mining perspective. We evaluate different supervised text classification algorithms on Contraint@AAAI 2021 Covid-19 Fake news detection dataset. The classification algorithms are based on Convolutional Neural Networks (CNN), Long Short Term Memory (LSTM), and Bidirectional Encoder Representations from Transformers (BERT). We also evaluate the importance of unsupervised learning in the form of language model pre-training and distributed word representations using unlabelled covid tweets corpus. We report the best accuracy of 98.41\% on the Covid-19 Fake news detection dataset.
Natural language processing (NLP) has been applied to various fields including text classification and sentiment analysis. In the shared task of sentiment analysis of code-mixed tweets, which is a part of the SemEval-2020 competition~\cite{patwa2020sentimix}, we preprocess datasets by replacing emoji and deleting uncommon characters and so on, and then fine-tune the Bidirectional Encoder Representation from Transformers(BERT) to perform the best. After exhausting top3 submissions, Our team MeisterMorxrc achieves an averaged F1 score of 0.730 in this task, and and our codalab username is MeisterMorxrc.
Here we proposed an approach to analyze text classification methods based on the presence or absence of task-specific terms (and their synonyms) in the text. We applied this approach to study six different transfer-learning and unsupervised methods for screening articles relevant to COVID-19 vaccines and therapeutics. The analysis revealed that while a BERT model trained on search-engine results generally performed well, it miss-classified relevant abstracts that did not contain task-specific terms. We used this insight to create a more effective unsupervised ensemble.
We study a family of data augmentation methods, substructure substitution (SUB2), for natural language processing (NLP) tasks. SUB2 generates new examples by substituting substructures (e.g., subtrees or subsequences) with ones with the same label, which can be applied to many structured NLP tasks such as part-of-speech tagging and parsing. For more general tasks (e.g., text classification) which do not have explicitly annotated substructures, we present variations of SUB2 based on constituency parse trees, introducing structure-aware data augmentation methods to general NLP tasks. For most cases, training with the augmented dataset by SUB2 achieves better performance than training with the original training set. Further experiments show that SUB2 has more consistent performance than other investigated augmentation methods, across different tasks and sizes of the seed dataset.
In this work, we introduce BanglaBERT, a BERT-based Natural Language Understanding (NLU) model pretrained in Bangla, a widely spoken yet low-resource language in the NLP literature. To pretrain BanglaBERT, we collect 27.5 GB of Bangla pretraining data (dubbed `Bangla2B+') by crawling 110 popular Bangla sites. We introduce two downstream task datasets on natural language inference and question answering and benchmark on four diverse NLU tasks covering text classification, sequence labeling, and span prediction. In the process, we bring them under the first-ever Bangla Language Understanding Benchmark (BLUB). BanglaBERT achieves state-of-the-art results outperforming multilingual and monolingual models. We are making the models, datasets, and a leaderboard publicly available at https://github.com/csebuetnlp/banglabert to advance Bangla NLP.
Noninformative uniform priors are staples of Bayesian inference, especially in Bayesian machine learning. This study challenges the assumption that they are optimal and their use in Bayesian inference yields optimal outcomes. Instead of using arbitrary noninformative uniform priors, we propose a machine learning based alternative method, learning optimal priors from data by maximizing a target function of interest. Applying na\"ive Bayes text classification methodology and a search algorithm developed for this study, our system learned priors from data using the positive predictive value metric as the target function. The task was to find Wikipedia articles that had not (but should have) been categorized under certain Wikipedia categories. We conducted five sets of experiments using separate Wikipedia categories. While the baseline models used the popular Bayes-Laplace priors, the study models learned the optimal priors for each set of experiments separately before using them. The results showed that the study models consistently outperformed the baseline models with a wide margin of statistical significance (p < 0.001). The measured performance improvement of the study model over the baseline was as high as 443% with the mean value of 193% over five Wikipedia categories.
The availability of large amounts of data and compelling computation power have made deep learning models much popular for text classification and sentiment analysis. Deep neural networks have achieved competitive performance on the above tasks when trained on naive text representations such as word count, term frequency, and binary matrix embeddings. However, many of the above representations result in the input space having a dimension of the order of the vocabulary size, which is enormous. This leads to a blow-up in the number of parameters to be learned, and the computational cost becomes infeasible when scaling to domains that require retaining a colossal vocabulary. This work proposes using singular value decomposition to transform the high dimensional input space to a lower-dimensional latent space. We show that neural networks trained on this lower-dimensional space are not only able to retain performance while savoring significant reduction in the computational complexity but, in many situations, also outperforms the classical neural networks trained on the native input space.
Pretrained language models (PLMs) perform poorly under adversarial attacks. To improve the adversarial robustness, adversarial data augmentation (ADA) has been widely adopted to cover more search space of adversarial attacks by adding textual adversarial examples during training. However, the number of adversarial examples for text augmentation is still extremely insufficient due to the exponentially large attack search space. In this work, we propose a simple and effective method to cover a much larger proportion of the attack search space, called Adversarial and Mixup Data Augmentation (AMDA). Specifically, AMDA linearly interpolates the representations of pairs of training samples to form new virtual samples, which are more abundant and diverse than the discrete text adversarial examples in conventional ADA. Moreover, to fairly evaluate the robustness of different models, we adopt a challenging evaluation setup, which generates a new set of adversarial examples targeting each model. In text classification experiments of BERT and RoBERTa, AMDA achieves significant robustness gains under two strong adversarial attacks and alleviates the performance degradation of ADA on the clean data. Our code is available at: https://github.com/thunlp/MixADA .
Transformers are not suited for processing long documents, due to their quadratically increasing memory and time consumption. Simply truncating a long document or applying the sparse attention mechanism will incur the context fragmentation problem or lead to an inferior modeling capability against comparable model sizes. In this paper, we propose ERNIE-Doc, a document-level language pretraining model based on Recurrence Transformers. Two well-designed techniques, namely the retrospective feed mechanism and the enhanced recurrence mechanism, enable ERNIE-Doc, which has a much longer effective context length, to capture the contextual information of a complete document. We pretrain ERNIE-Doc to explicitly learn the relationships among segments with an additional document-aware segment-reordering objective. Various experiments were conducted on both English and Chinese document-level tasks. ERNIE-Doc improved the state-of-the-art language modeling result of perplexity to 16.8 on WikiText-103. Moreover, it outperformed competitive pretraining models by a large margin on most language understanding tasks, such as text classification and question answering.
For both human readers and pre-trained language models (PrLMs), lexical diversity may lead to confusion and inaccuracy when understanding the underlying semantic meanings of given sentences. By substituting complex words with simple alternatives, lexical simplification (LS) is a recognized method to reduce such lexical diversity, and therefore to improve the understandability of sentences. In this paper, we leverage LS and propose a novel approach which can effectively improve the performance of PrLMs in text classification. A rule-based simplification process is applied to a given sentence. PrLMs are encouraged to predict the real label of the given sentence with auxiliary inputs from the simplified version. Using strong PrLMs (BERT and ELECTRA) as baselines, our approach can still further improve the performance in various text classification tasks.
We study an important and challenging task of attacking natural language processing models in a hard label black box setting. We propose a decision-based attack strategy that crafts high quality adversarial examples on text classification and entailment tasks. Our proposed attack strategy leverages population-based optimization algorithm to craft plausible and semantically similar adversarial examples by observing only the top label predicted by the target model. At each iteration, the optimization procedure allow word replacements that maximizes the overall semantic similarity between the original and the adversarial text. Further, our approach does not rely on using substitute models or any kind of training data. We demonstrate the efficacy of our proposed approach through extensive experimentation and ablation studies on five state-of-the-art target models across seven benchmark datasets. In comparison to attacks proposed in prior literature, we are able to achieve a higher success rate with lower word perturbation percentage that too in a highly restricted setting.
In artificial intelligence (AI), knowledge is the information required by an intelligent system to accomplish tasks. While traditional knowledge bases use discrete, symbolic representations, detecting knowledge encoded in the continuous representations learned from data has received increasing attention recently. In this work, we propose a method for building a continuous knowledge base (CKB) that can store knowledge imported from multiple, diverse neural networks. The key idea of our approach is to define an interface for each neural network and cast knowledge transferring as a function simulation problem. Experiments on text classification show promising results: the CKB imports knowledge from a single model and then exports the knowledge to a new model, achieving comparable performance with the original model. More interesting, we import the knowledge from multiple models to the knowledge base, from which the fused knowledge is exported back to a single model, achieving a higher accuracy than the original model. With the CKB, it is also easy to achieve knowledge distillation and transfer learning. Our work opens the door to building a universal continuous knowledge base to collect, store, and organize all continuous knowledge encoded in various neural networks trained for different AI tasks.
We study an important task of attacking natural language processing models in a black box setting. We propose an attack strategy that crafts semantically similar adversarial examples on text classification and entailment tasks. Our proposed attack finds candidate words by considering the information of both the original word and its surrounding context. It jointly leverages masked language modelling and next sentence prediction for context understanding. In comparison to attacks proposed in prior literature, we are able to generate high quality adversarial examples that do significantly better both in terms of success rate and word perturbation percentage.
The state of the art in learning meaningful semantic representations of words is the Transformer model and its attention mechanisms. Simply put, the attention mechanisms learn to attend to specific parts of the input dispensing recurrence and convolutions. While some of the learned attention heads have been found to play linguistically interpretable roles, they can be redundant or prone to errors. We propose a method to guide the attention heads towards roles identified in prior work as important. We do this by defining role-specific masks to constrain the heads to attend to specific parts of the input, such that different heads are designed to play different roles. Experiments on text classification and machine translation using 7 different datasets show that our method outperforms competitive attention-based, CNN, and RNN baselines.
Humor recognition has been widely studied as a text classification problem using data-driven approaches. However, most existing work does not examine the actual joke mechanism to understand humor. We break down any joke into two distinct components: the set-up and the punchline, and further explore the special relationship between them. Inspired by the incongruity theory of humor, we model the set-up as the part developing semantic uncertainty, and the punchline disrupting audience expectations. With increasingly powerful language models, we were able to feed the set-up along with the punchline into the GPT-2 language model, and calculate the uncertainty and surprisal values of the jokes. By conducting experiments on the SemEval 2021 Task 7 dataset, we found that these two features have better capabilities of telling jokes from non-jokes, compared with existing baselines.
Providing pretrained language models with simple task descriptions in natural language enables them to solve some tasks in a fully unsupervised fashion. Moreover, when combined with regular learning from examples, this idea yields impressive few-shot results for a wide range of text classification tasks. It is also a promising direction to improve data efficiency in generative settings, but there are several challenges to using a combination of task descriptions and example-based learning for text generation. In particular, it is crucial to find task descriptions that are easy to understand for the pretrained model and to ensure that it actually makes good use of them; furthermore, effective measures against overfitting have to be implemented. In this paper, we show how these challenges can be tackled: We introduce GenPET, a method for text generation that is based on pattern-exploiting training, a recent approach for combining textual instructions with supervised learning that only works for classification tasks. On several summarization and headline generation datasets, GenPET gives consistent improvements over strong baselines in few-shot settings.
This paper illustrates the details description of technical text classification system and its results that developed as a part of participation in the shared task TechDofication 2020. The shared task consists of two sub-tasks: (i) first task identify the coarse-grained technical domain of given text in a specified language and (ii) the second task classify a text of computer science domain into fine-grained sub-domains. A classification system (called 'TechTexC') is developed to perform the classification task using three techniques: convolution neural network (CNN), bidirectional long short term memory (BiLSTM) network, and combined CNN with BiLSTM. Results show that CNN with BiLSTM model outperforms the other techniques concerning task-1 of sub-tasks (a, b, c and g) and task-2a. This combined model obtained f1 scores of 82.63 (sub-task a), 81.95 (sub-task b), 82.39 (sub-task c), 84.37 (sub-task g), and 67.44 (task-2a) on the development dataset. Moreover, in the case of test set, the combined CNN with BiLSTM approach achieved that higher accuracy for the subtasks 1a (70.76%), 1b (79.97%), 1c (65.45%), 1g (49.23%) and 2a (70.14%).