6 research outputs found
Cross-Linguistic Offensive Language Detection: BERT-Based Analysis of Bengali, Assamese, & Bodo Conversational Hateful Content from Social Media
Get PDFAbstract
In today’s age, social media reigns as the paramount communication platform, providing individuals with the avenue to express their conjectures, intellectual propositions, and reflections. Unfortunately, this freedom often comes with a downside as it facilitates the widespread proliferation of hate speech and offensive content, leaving a deleterious impact on our world. Thus, it becomes essential to discern and eradicate such offensive material from the realm of social media. This article delves into the comprehensive results and key revelations from the HASOC-2023 offensive language identification result. The primary emphasis is placed on the meticulous detection of hate speech within the linguistic domains of Bengali, Assamese, and Bodo, forming the framework for Task 4: Annihilate Hates. In this work, we used BERT models, including XML-Roberta, L3-cube, IndicBERT, BenglaBERT, and BanglaHateBERT. The research outcomes were promising and showed that XML-Roberta-lagre performed better than monolingual models in most cases. Our team ’TeamBD’ achieved rank 3rd for Task 4 - Assamese, & 5th for Bengali.Abstract
In today’s age, social media reigns as the paramount communication platform, providing individuals with the avenue to express their conjectures, intellectual propositions, and reflections. Unfortunately, this freedom often comes with a downside as it facilitates the widespread proliferation of hate speech and offensive content, leaving a deleterious impact on our world. Thus, it becomes essential to discern and eradicate such offensive material from the realm of social media. This article delves into the comprehensive results and key revelations from the HASOC-2023 offensive language identification result. The primary emphasis is placed on the meticulous detection of hate speech within the linguistic domains of Bengali, Assamese, and Bodo, forming the framework for Task 4: Annihilate Hates. In this work, we used BERT models, including XML-Roberta, L3-cube, IndicBERT, BenglaBERT, and BanglaHateBERT. The research outcomes were promising and showed that XML-Roberta-lagre performed better than monolingual models in most cases. Our team ’TeamBD’ achieved rank 3rd for Task 4 - Assamese, & 5th for Bengali
Offensive Language Identification Using Hindi-English Code-Mixed Tweets, and Code-Mixed Data Augmentation
Get PDFAbstractThe Code-mixed text classification is challenging due to the lack of code-mixed labeled datasets and the non-existence of pre-trained models. This paper presents the HASOC-2021 offensive language identification results and main findings on code-mixed (Hindi-English) Subtask2. In this work, we have proposed a new method of code-mixed data augmentation using synonym replacement of Hindi and English words using WordNet, and phonetics conversion of Hinglish (Hindi-English) words. We used a 5.7k pre-annotated HASOC-2021 code-mixed dataset for training and data augmentation. The proposal’s feasibility was tested with a Logistic Regression (LR) used as a baseline, Convolutional Neural Network (CNN), and BERT with and without data augmentation. The research outcomes were promising and yields almost 3% increase of classifier accuracy and F1 scores as compared to baseline. Our official submission showed a 66.56% F1 score and ranked 8th position in the competition.Abstract
The Code-mixed text classification is challenging due to the lack of code-mixed labeled datasets and the non-existence of pre-trained models. This paper presents the HASOC-2021 offensive language identification results and main findings on code-mixed (Hindi-English) Subtask2. In this work, we have proposed a new method of code-mixed data augmentation using synonym replacement of Hindi and English words using WordNet, and phonetics conversion of Hinglish (Hindi-English) words. We used a 5.7k pre-annotated HASOC-2021 code-mixed dataset for training and data augmentation. The proposal’s feasibility was tested with a Logistic Regression (LR) used as a baseline, Convolutional Neural Network (CNN), and BERT with and without data augmentation. The research outcomes were promising and yields almost 3% increase of classifier accuracy and F1 scores as compared to baseline. Our official submission showed a 66.56% F1 score and ranked 8th position in the competition
Offensive language identification using Hindi-English code-mixed tweets, and code-mixed data augmentation
No full textAbstract
The Code-mixed text classification is challenging due to the lack of code-mixed labeled datasets and the non-existence of pre-trained models. This paper presents the HASOC-2021 offensive language identification results and main findings on code-mixed (Hindi-English) Subtask2. In this work, we have proposed a new method of code-mixed data augmentation using synonym replacement of Hindi and English words using WordNet, and phonetics conversion of Hinglish (Hindi-English) words. We used a 5.7k pre-annotated HASOC-2021 code-mixed dataset for training and data augmentation. The proposal’s feasibility was tested with a Logistic Regression (LR) used as a baseline, Convolutional Neural Network (CNN), and BERT with and without data augmentation. The research outcomes were promising and yields almost 3% increase of classifier accuracy and F1 scores as compared to baseline. Our official submission showed a 66.56% F1 score and ranked 8th position in the competition
A Comprehensive Study on NLP Data Augmentation for Hate Speech Detection: Legacy Methods, BERT, and LLMs
Get PDFThe surge of interest in data augmentation within the realm of NLP has been
driven by the need to address challenges posed by hate speech domains, the
dynamic nature of social media vocabulary, and the demands for large-scale
neural networks requiring extensive training data. However, the prevalent use
of lexical substitution in data augmentation has raised concerns, as it may
inadvertently alter the intended meaning, thereby impacting the efficacy of
supervised machine learning models. In pursuit of suitable data augmentation
methods, this study explores both established legacy approaches and
contemporary practices such as Large Language Models (LLM), including GPT in
Hate Speech detection. Additionally, we propose an optimized utilization of
BERT-based encoder models with contextual cosine similarity filtration,
exposing significant limitations in prior synonym substitution methods. Our
comparative analysis encompasses five popular augmentation techniques: WordNet
and Fast-Text synonym replacement, Back-translation, BERT-mask contextual
augmentation, and LLM. Our analysis across five benchmarked datasets revealed
that while traditional methods like back-translation show low label alteration
rates (0.3-1.5%), and BERT-based contextual synonym replacement offers sentence
diversity but at the cost of higher label alteration rates (over 6%). Our
proposed BERT-based contextual cosine similarity filtration markedly reduced
label alteration to just 0.05%, demonstrating its efficacy in 0.7% higher F1
performance. However, augmenting data with GPT-3 not only avoided overfitting
with up to sevenfold data increase but also improved embedding space coverage
by 15% and classification F1 score by 1.4% over traditional methods, and by
0.8% over our method.Comment: 31 page, Table 18, Figure 1
Road Condition Detection and Crowdsourced Data Collection for Accident Prevention: A Deep Learning Approach
Get PDFAbstract
Bangladesh is one of the countries struggling to prevent road accidents, which is a global cause for concern. An early warning system that indicates road conditions can contribute to the prevention task. For this purpose, a deep-learning based approach using a Convolutional Neural Network (CNN) to learn from random road images the safety factor is developed. This results in a three-class categorization: (i) Severely risky roads, (ii) Mildly risky roads, and (iii) Normal roads. The application of deep learning techniques in this study yields an accuracy of 95.5% in detecting problematic road conditions. Furthermore, based on the study’s findings, a mobile application has been developed. The app enables real-time crowdsourced data collection of road conditions and provides a platform for users to share this information in real-time with other drivers, thereby, contributing to prevent accidents and raise awareness among drivers and users by pinpointing the location of the risky road. Finally, crowdsourced data has been reused to update the trained model, which further improves the classifier accuracy.Abstract
Bangladesh is one of the countries struggling to prevent road accidents, which is a global cause for concern. An early warning system that indicates road conditions can contribute to the prevention task. For this purpose, a deep-learning based approach using a Convolutional Neural Network (CNN) to learn from random road images the safety factor is developed. This results in a three-class categorization: (i) Severely risky roads, (ii) Mildly risky roads, and (iii) Normal roads. The application of deep learning techniques in this study yields an accuracy of 95.5% in detecting problematic road conditions. Furthermore, based on the study’s findings, a mobile application has been developed. The app enables real-time crowdsourced data collection of road conditions and provides a platform for users to share this information in real-time with other drivers, thereby, contributing to prevent accidents and raise awareness among drivers and users by pinpointing the location of the risky road. Finally, crowdsourced data has been reused to update the trained model, which further improves the classifier accuracy
A Meandered Line Patch Antenna at Low Frequency Range for Early Stage Breast Cancer Detection
Get PDFEvery year a concerning number of women are affected by breast cancer which is one of the deadliest and common types of cancers. Breast cancer is curable at early stages. For detecting breast cancer, there are several methods such as MRI, Mammography, Tomography, Ultrasound, and biopsy are available in medical technology. Still, none of them are as easy and efficient as a microwave imaging technique, in this method, the antenna plays an important role. Therefore, this paper focuses on developing an antenna at a low-frequency range for microwave imaging techniques to detect cancerous tissue inside the breast. For this, the antenna parameters, i.e., return loss, VSWR, directivity, current density, and specific absorption rate were studied, by setting the antenna over without tumor and with tumor breast as up-side-down, to ensure the compatibility of the antenna for the technique as well as for the patient’s body. A 5mm radius cancerous tumor was created inside the breast with dielectric conductivity of 4 and relative permittivity of 50. Cancerous cells were detected by reading the antenna parameters’ comparison between the healthy breast and the affected breast. The whole study was conducted by using CST MICROWAVE STUDIO SUITE 2020.
