Bangla Text Research Articles

Punctuation Prediction in Bangla Text

Punctuation prediction is critical as it can enhance the readability of machine-transcribed speeches or texts significantly by adding appropriate punctuation. Furthermore, systems like Automatic Speech Recognizer (ASR) produce texts that are unpunctuated, making the readability difficult for humans and also hampers the performance of various natural language processing (NLP) tasks. Such NLP related tasks have been investigated thoroughly for English; however, very limited work is done for punctuation prediction in the Bangla language. In this study, we train a bidirectional recurrent neural network (BRNN) along with Attention model with a plausibly large Bangla dataset. Afterwards, we apply extensive postprocessing techniques for predicting punctuation more accurately with the employed model. Initially, we perform experimentation with a relatively imbalanced dataset, and our model shows promising results F1=56.9 for Period) in punctuation prediction. Later, we also investigate the model’s performance using a balanced Bangla dataset to achieve higher performance scores ( F1=62.2 for Question). Thus, the goal of this study is to propose an efficient approach that can predict punctuation in Bangla texts effectively. Our study also includes investigation on how our postprocessing techniques affect the prediction performance. Being an early attempt for the punctuation prediction in Bangla text, our work is expected to significantly contribute in the NLP field for the Bangla language, and will pave the way for future work with the Bangla language in this direction.

Multi-class sentiment classification on Bengali social media comments using machine learning

Multi-class Sentiment Analysis (SA) is an important field of computational linguistics that extracts multiple opinions expressed in a text using NLP and text-mining techniques. Existing research on multi-class SA in the Bengali language is directed towards ternary classification with unsatisfactory classification performance. Moreover, obtaining a higher performance score is challenging due to the peculiarities of Bengali text, lack of ground truth datasets, and low resources of preprocessing tools. Moreover, no research has shown that deep learning algorithms perform higher on four types of sentiments. Therefore, we proposed a supervised deep learning classifier based on CNN and LSTM to conduct multi-class SA on Bengali social media comments labelled as sexual, religious, political, and acceptable. The study aims to achieve maximum accuracy using the proposed model and provide a comparative analysis with the baseline models. Six machine learning models with two different feature extraction techniques were considered baseline models. The performance of our proposed CLSTM architecture can greatly improve the performance of SA with 85.8% accuracy and 0.86 F1 scores on a labelled dataset of 42,036 Facebook comments. A web application based on the proposed model and the highest-performing baseline model was built to detect the real-life sentiment of social media comments.

Social Media Cyberbullying Detection on Political Violence from Bangla Texts Using Machine Learning Algorithm

Social Media Cyberbullying Detection on Political Violence from Bangla Texts Using Machine Learning Algorithm

Text Recognition of Bangla and English Scripts in Natural Scene Images

Digital cameras on mobile devices have created several new computational hurdles. Text extraction from natural scene photos collected by such devices is an issue. This paper presents a method for recognizing English and Bangla text in scene photos. With this framework, it will be possible to recognize and locate text in its natural setting. Text recognition in natural settings is performed on multiple distinct tiers. To begin, a picture is taken using a smart device. Then text localization and text extraction are done. In the final stage, an artificial neural network is used to recognize the text from the image. This is a bilingual project, with results available in both Bengali and English. The system outperforms conventional recognition methods, producing image results with an average accuracy of 85%.

Deep Learning Based Misogynistic Bangla Text Identification from Social Media

Deep Learning Based Misogynistic Bangla Text Identification from Social Media

SECURING BANGLA TEXT COMMUNICATION USING IMAGE STEGANOGRAPHY WITH DYNAMIC SUBSTITUTION IN IOT ENVIRONMENT

Privacy and security of information are the prime concerns of today’s internet users. They want to get better and reliable services without sacrificing privacy from the current paradigm like the Internet of Things (IoT). In order to ensure IoT as a truthful service platform for the vast users, we need to protect the information that evolves throughout the internet and in the storage. In this paper, we propose a steganographic method for Bangla text communication over the unsecured internet based on RGB color image with the help of secret key for better protection of information. The proposed system consists of four main components namely preprocessing, embedding, extraction, and post-processing. We use dynamic positioning in case of substitution of (Bangla) text into the image. The secret key is transmitting throughput the image so that there is no extra hassle for communicating the secret key. We justify our proposed approach by using simulation with respect to imperceptibility, capacity, and robustness. We compare the result of the proposed method with other existing methods and get a better result over several existing efficient methods.

An attention-based hybrid architecture with explainability for depressive social media text detection in Bangla

Mental health has become a major concern in recent years. Social media have been increasingly used as platforms to gain insight into a person’s mental health condition by analysing the posts and comments, which are textual in nature. By analysing these texts, depressive posts can be detected. To facilitate this process, this work presents an attention-based bidirectional Long Short-Term Memory (LSTM)- Convolutional Neural Network (CNN) based model to detect depressive Bangla social media texts, which is lighter and more robust than the conventional models and provides better performance. A dataset containing such Bangla texts was also developed in this work to mitigate the scarcity. Different preprocessing stages were followed, and three embeddings were used in this task. Thanks to the attention mechanism, the proposed model achieved an accuracy of 94.3% with 92.63% of sensitivity and 95.12% of specificity. When tested on other languages, such as English, the proposed model performed remarkably. The robustness and explainability of the proposed model were also discussed in this paper. Additionally, when compared with classical machine learning models, ensemble approaches, transformers, other similar models, and existing architectures, the proposed model outperformed them.

A hybrid convolutional long short-term memory (CNN-LSTM) based natural language processing (NLP) model for sentiment analysis of customer product reviews in Bangla

Sentiment Analysis (SA) examines how people feel about products, services, people, organizations, events etc. Most of the Natural language processing on SA research has focused on English. In the case of Bangla, it lacks sufficient study as well as a proper dataset. All previous analyses employed KNN, NB, and other methods. We develop a natural language processing (NLP) model for separating opinion and sentiment from Bangla customer surveys. This technology isolates extreme client opinions to help with business and marketing decisions. Bangladesh is embracing e-commerce and f-commerce. Client comments and evaluations are becoming more significant in judging product or service quality, and this industry is evolving toward internet distribution. Organizations utilize client audits to check product quality. Our objective is to systematically collect client feedback and understand their product reaction. We used a hybrid CNN-LSTM based NLP model to classify Bangla texts in 3 viewpoint categories (positive, negative and neutral). We tested our model using a Bangla dataset that we generated. For our dataset, we collected polls and comments from websites and social media. Finally, among the three evaluation matrices, the f-1 score is providing the highest average, and the three-opinion technique is 87.22 percent accurate in determining the performance of our task.

Detection of Depression Severity Using Bengali Social Media Posts on Mental Health: Study Using Natural Language Processing Techniques.

BackgroundThere are a myriad of language cues that indicate depression in written texts, and natural language processing (NLP) researchers have proven the ability of machine learning and deep learning approaches to detect these cues. However, to date, these approaches bridging NLP and the domain of mental health for Bengali literature are not comprehensive. The Bengali-speaking population can express emotions in their native language in greater detail.ObjectiveOur goal is to detect the severity of depression using Bengali texts by generating a novel Bengali corpus of depressive posts. We collaborated with mental health experts to generate a clinically sound labeling scheme and an annotated corpus to train machine learning and deep learning models.MethodsWe conducted a study using Bengali text-based data from blogs and open source platforms. We constructed a procedure for annotated corpus generation and extraction of textual information from Bengali literature for predictive analysis. We developed our own structured data set and designed a clinically sound labeling scheme with the help of mental health professionals, adhering to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) during the process. We used 5 machine learning models for detecting the severity of depression: kernel support vector machine (SVM), random forest, logistic regression K-nearest neighbor (KNN), and complement naive Bayes (NB). For the deep learning approach, we used long short-term memory (LSTM) units and gated recurrent units (GRUs) coupled with convolutional blocks or self-attention layers. Finally, we aimed for enhanced outcomes by using state-of-the-art pretrained language models.ResultsThe independent recurrent neural network (RNN) models yielded the highest accuracies and weighted F1 scores. GRUs, in particular, produced 81% accuracy. The hybrid architectures could not surpass the RNNs in terms of performance. Kernel SVM with term frequency–inverse document frequency (TF-IDF) embeddings generated 78% accuracy on test data. We used validation and training loss curves to observe and report the performance of our architectures. Overall, the number of available data remained the limitation of our experiment.ConclusionsThe findings from our experimental setup indicate that machine learning and deep learning models are fairly capable of assessing the severity of mental health issues from texts. For the future, we suggest more research endeavors to increase the volume of Bengali text data, in particular, so that modern architectures reach improved generalization capability.

A Dynamic Strategy for Classifying Sentiment From Bengali Text by Utilizing Word2vector Model

In today's world, around 230 million people used the Bengali or Bangla language to communicate. These individuals are progressively associated with online exercises on famous micro-blogging and long-range interpersonal communication locales, imparting insights what's more, musings, and also the vast majority of articles are in the Bengali language. Thus, Bengali people express their emotions using the Bangla language by reviewing, commenting, or recommendations. Sentiment analysis helps determine the people's emotions expressed on social media or several online platforms. Therefore, this study focused on extracting their emotion from a Bengali text by utilizing Word2vector, Skip-Gram, and Continuous Bag of Words (CBOW) with a new Word to Index model by focusing on three individual classes happy, angry, and excited. The authors achieved the highest accuracy of 75% by utilizing the skip-gram model to classify those three types of emotions. This study also outperformed other existing works with LSTM, CNN model with existing datasets.

How can we manage Offensive Text in Social Media - A Text Classification Approach using LSTM-BOOST

Recently, offensive content has become increasingly popular for harassing and criticizing people on numerous social media platforms. This paper proposes an offensive text classification algorithm named LSTM-BOOST employing Long Short-Term Memory(LSTM) model with ensemble learning to recognize offensive Bengali texts in various social media platforms. The proposed LSTM-BOOST model uses the modified AdaBoost algorithm employing principal component analysis(PCA) along with LSTM networks. In the LSTM-Boost model, the dataset is divided into three categories, and PCA and LSTM networks are applied to each part of the dataset to obtain the most significant variance and reduce the weighted error of the weak hypothesis of the model. Furthermore, different classifiers are used for baseline experiment and the model is evaluated on various word embedding vector methods. Our investigation found that the LSTM-BOOST algorithms outperform most of the baseline architecture, leading F1-score of 92.61% on the Bengali offensive text from Social Platforms(BHSSP) dataset.

Bangla text generation system by incorporating attention in sequence-to-sequence model

In this AI-driven digital era, the pervasive nature of digital data is possible due to the widespread and cheap access to the Internet. Internet is continuously flourishing with data in many forms. Among them, textual data are a great source of information where people share their expressions in written format. Social media, blogs, online newspapers, government documents are some notable textual data sources. Information extraction from this enormous amount of data by manual inspection is time-consuming, cumbersome, and sometimes impossible. Natural Language Processing (NLP) is the computational domain for addressing those limitations by solving human language-related problems. Text summarization, Named entity recognition, Question answering are some of them where a common task for a machine is to generate coherent text. In such scenarios, the input is a sequence of text, and the output is also a sequence, but they differ in length. Sequence-to-Sequence (Seq2Seq) is an algorithmic approach to address that scenario by utilizing layers of recurrent units. However, the simple Seq2Seq model fails to capture the long-term relationship on the input sequence. Research shows that the attention mechanism guides the model to concentrate on specific inputs. Existing literature shows a lack of quality research on this text generation problem in the Bangla language, whereas many languages show excellent results. This work aims to develop such a system by incorporating attention to the Seq2Seq model and justifying its applicability by comparing it with baseline models. The model perplexity shows that the system can generate human-level readable text using a preprocessed dataset.

Bangla text generation system by incorporating attention in sequence-to-sequence model

Bangla text generation system by incorporating attention in sequence-to-sequence model

Breaking the Curse of Class Imbalance: Bangla Text Classification

This article addresses the class imbalance issue in a low-resource language called Bengali. As a use-case, we choose one of the most fundamental NLP tasks, i.e., text classification, where we utilize three benchmark text corpora: fake-news dataset, sentiment analysis dataset, and song lyrics dataset. Each of them contains a critical class imbalance. We attempt to tackle the problem by applying several strategies that include data augmentation with synthetic samples via text and embedding generation in order to augment the proportion of the minority samples. Moreover, we apply ensembling of deep learning models by subsetting the majority samples. Additionally, we enforce the focal loss function for class-imbalanced data classification. We also apply the outlier detection technique, data resampling, and hidden feature extraction to improve the minority-f1 score. All of our experimentations are entirely focused on textual content analysis, which results in a more than90%minority f1 score for each of the three tasks. It is an excellent outcome on such highly class-imbalanced datasets.

Translanguaging as a Transformative Act in a Reading Classroom: Perspectives from a Bangladeshi Private University

ABSTRACT This study investigated the role of translanguaging pedagogy in reading comprehension of first-year students studying in an English medium classroom of a Bangladeshi private university. Data were collected through classroom observations, pedagogical intervention, focus group discussion with students and a semi-structured interview with the class teacher. The notions of ”translanguaging space” and ”critical instances” were used to analyse the data. The results show that the translanguaging space provided scope in each step of the intervention to maximise the use of the linguistic and semiotic resources of students, putting them at ease, and enhancing epistemic access to and comprehension of complicated English texts. The purposeful design of the lecture, including scaffolding with multilingual words and expressions, and guided reading with Bangla text and topics pertaining to students’ lives, engaged them deeply with the content, while also transforming knowledge and subjectivities about the given topic.

An efficient ROI detection algorithm for Bangla text extraction and recognition from natural scene images

This research work plays a significant role in finding information from the scene images to fulfill the demand of real life applications like detection of license plate, navigation of robot and helping the visually impaired persons. Here, a new algorithm has been proposed and applied on the scene images to extract Region of Interest (ROI). All the Bangla words are then separated from a sentence by analyzing and applying the Connected Component (CC) method along with bounding box technology. Another new algorithm has been proposed and applied to apart and bring-out Bangla characters from the Bangla words. This algorithm works by the method of vertical scanning of the images of Bangla words. Finally, the extracted characters are recognized by using the Support Vector Machine (SVM) as a classifier which works with Histogram of Oriented Gradient (HOG) features. There are 500 scene images with variations in colors, writing styles and orientations in our designed database. The proposed algorithm yields the accuracy 92.70% and 93.23% in extraction of ROI and character respectively. In the recognition of Bangla characters (digits, Basic characters, and joined characters), the average accuracy is 99.16%. The recognition accuracy of Bangla characters using Convolutional Neural Network (CNN) is also calculated and the obtained result is 83.52%.

Sentiment analysis on Bangla text using extended lexicon dictionary and deep learning algorithms

Sentiment analysis (SA) is a subset of natural language processing (NLP) research. In the case of categorical weighted based dictionary with rule-based sentiment score generation, no work in SA has been done yet in Bangla language using deep learning (DL) approaches. This paper proposes DL models for SA on Bangla text using an extended lexicon data dictionary (LDD). We implement the rule-based method Bangla text sentiment score (BTSC) algorithm for extracting polarity from large texts. These polarities are then fed into the neural network along with the preprocessed text as training samples. The preprocessed texts are formatted as a vectorization of words of unique numbers of pre-trained word embedding models. Word2Vec matrix with top highest probability word is applied on embedding layer as a weighted matrix to fit the DL models. This paper also presents a remarkably detailed analysis of selective DL models with some fine tuning. The fine-tuning includes the use of drop out, optimizer regularization, learning rate, multiple layers, filters, attention mechanism, capsule layers, transformer with progressive training along with validation and testing accuracy, precision, recall and F1-score. Experimental results indicate that the proposed new long short-term memory (LSTM) models are highly accurate in performing SA tasks. For our proposed hierarchical attention based LSTM (HAN-LSTM), Dynamic routing based capsule neural network with Bi-LSTM (D-CAPSNET-Bi-LSTM) and bidirectional encoder representations from Transformers (BERT) with LSTM (BERT-LSTM) model we achieved accuracy values of 78.52%, 80.82% and 84.18% respectively.

BEmoC: A Corpus for Identifying Emotion in Bengali Texts.

Emotion classification in text has growing interest among NLP experts due to the enormous availability of people’s emotions and its emergence on various Web 2.0 applications/services. Emotion classification in the Bengali texts is also gradually being considered as an important task for sports, e-commerce, entertainments, and security applications. However, It is a very critical task to develop an automatic emotion classification system for low-resource languages such as, Bengali. Scarcity of resources and deficiency of benchmark corpora make the task more complicated. Thus, the development of a benchmark corpus is the prerequisite to develop an emotion classifier for Bengali texts. This paper describes the development of an emotional corpus (hereafter called ‘BEmoC’) for classifying six emotions in Bengali texts. The corpus development process consists of four key steps: data crawling, pre-processing, labelling, and verification. A total of 7000 texts are labelled into six basic emotion categories such as anger, fear, surprise, sadness, joy, and disgust, respectively. Dataset evaluation with 0.969 Cohen’s κ score indicates the close agreement between the corpus annotators and the expert. The analysis of evaluation also represents that the distribution of emotion words obeys Zipf’s law. Moreover, the results of BEmoC analysis shown in terms of coding reliability, emotion density, and most frequent emotion words, respectively.

Emotion Detection from Multilingual Text and Multi-Emotional Sentence using Difference NLP Feature Extraction Technique and ML Classifier

Machines can read, comprehend, and extrapolate meaning from human languages, thanks to natural language processing.In this paper, we have detected emotion from multilingual text and multi-emotional sentences.For our research, we have collected a dataset containing around 7000 tweets on 4 emotions (Anger, Fear, Joy, and Sadness). After pre-processing our data, we used 2 NLP feature extraction models and trained those with the help of 4 different Machine Learning classifiers. We have also developed an algorithm for detecting exact emotions from multi-emotional sentences. Also, we compared our result with a research paper using the same dataset (ISEAR). And found out our model provides relatively better resultsthan that model. We also tried to determine emotion from the Bangla text. Although there is not much data regarding emotion in Bengali. We managed to get around 600 data on Bangla.

Bengali paper classification using ensemble machine learning algorithms

Text classification is one of the most challenging problems in natural language processing (NLP). Language models are at the heart of NLP. The ability to represent texts as numbers has given rise to many NLP tasks, for example, text categorisation, translation, and summarisation. Unfortunately, NLP for Bengali texts has not reached the state-of-art level of other Languages like English yet, mostly due to the scarcity of resources and the complexities seen in Bengali grammar. Therefore, not much work has been done in this field. In this paper, we have studied one of the word embedding methods, Word2vec, based on continuous bag of words (CBOW) with several ensemble machine learning algorithms, e.g., Adaptive Boosting Classifiers, Light Gradient Boosting Machine, XGboost, and random forest classifiers (RFC). The model is trained on a large corpus of Bengali newspapers of a considerable size that has 99283949 words and 8284804 sentences in 392772 documents. In our experiment, Word2vec CBOW model with XGboost algorithm performed much better than other models and achieved 92.24% accuracy.