Search

The prevalence of cyberbullying has reached an alarming rate, affecting approximately 54% of teenagers who experience various forms of cyberbullying, including offensive hate speech, threats, and racism. This research introduces a comprehensive dataset and system for cyberbullying detection in Urdu tweets, leveraging a spectrum of machine learning approaches including traditional models and advanced deep learning techniques. The objectives of this study are threefold. Firstly, a dataset consisting of 12,500 annotated tweets in Urdu is created, and it is made publicly available to the research community. Secondly, annotation guidelines for Urdu text with appropriate labels for cyberbullying detection are developed. Finally, a series of experiments is conducted to assess the performance of machine learning and deep learning techniques in detecting cyberbullying. The results indicate that fastText deep learning models outperform other models in cyberbullying detection. This study demonstrates its efficacy in effectively detecting and classifying cyberbullying incidents in Urdu tweets, contributing to the broader effort of creating a safer digital environment. [ABSTRACT FROM AUTHOR]

Sentiment analysis, a crucial task in discerning emotional tones within the text, plays a pivotal role in understanding public opinion and user sentiment across diverse languages. While numerous scholars conduct sentiment analysis in widely spoken languages such as English, Chinese, Arabic, Roman Arabic, and more, we come to grappling with resource-poor languages like Urdu literature which becomes a challenge. Urdu is a uniquely crafted language, characterized by a script that amalgamates elements from diverse languages, including Arabic, Parsi, Pashtu, Turkish, Punjabi, Saraiki, and more. As Urdu literature, characterized by distinct character sets and linguistic features, presents an additional hurdle due to the lack of accessible datasets, rendering sentiment analysis a formidable undertaking. The limited availability of resources has fueled increased interest among researchers, prompting a deeper exploration into Urdu sentiment analysis. This research is dedicated to Urdu language sentiment analysis, employing sophisticated deep learning models on an extensive dataset categorized into five labels: Positive, Negative, Neutral, Mixed, and Ambiguous. The primary objective is to discern sentiments and emotions within the Urdu language, despite the absence of well-curated datasets. To tackle this challenge, the initial step involves the creation of a comprehensive Urdu dataset by aggregating data from various sources such as newspapers, articles, and socialmedia comments. Subsequent to this data collection, a thorough process of cleaning and preprocessing is implemented to ensure the quality of the data. The study leverages two well-known deep learningmodels, namely Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), for both training and evaluating sentiment analysis performance. Additionally, the study explores hyperparameter tuning to optimize the models' efficacy. Evaluation metrics such as precision, recall, and the F1-score are employed to assess the effectiveness of the models. The research findings reveal that RNN surpasses CNN in Urdu sentiment analysis, gaining a significantly higher accuracy rate of 91%. This result accentuates the exceptional performance of RNN, solidifying its status as a compelling option for conducting sentiment analysis tasks in the Urdu language. [ABSTRACT FROM AUTHOR]

The prevalence of cyberbullying has reached an alarming rate, affecting approximately 54% of teenagers who experience various forms of cyberbullying, including offensive hate speech, threats, and racism. This research introduces a comprehensive dataset and system for cyberbullying detection in Urdu tweets, leveraging a spectrum of machine learning approaches including traditional models and advanced deep learning techniques. The objectives of this study are threefold. Firstly, a dataset consisting of 12,500 annotated tweets in Urdu is created, and it is made publicly available to the research community. Secondly, annotation guidelines for Urdu text with appropriate labels for cyberbullying detection are developed. Finally, a series of experiments is conducted to assess the performance of machine learning and deep learning techniques in detecting cyberbullying. The results indicate that fastText deep learning models outperform other models in cyberbullying detection. This study demonstrates its efficacy in effectively detecting and classifying cyberbullying incidents in Urdu tweets, contributing to the broader effort of creating a safer digital environment.

Background/introduction: The dawn of the internet opened the doors to the easy and widespread sharing of information on subject matters such as products, services, events and political opinions. While the volume of studies conducted on sentiment analysis is rapidly expanding, these studies mostly address English language concerns. The primary goal of this study is to present state-of-art survey for identifying the progress and shortcomings saddling Urdu sentiment analysis and propose rectifications. Methods: We described the advancements made thus far in this area by categorising the studies along three dimensions, namely: text pre-processing lexical resources and sentiment classification. These pre-processing operations include word segmentation, text cleaning, spell checking and part-of-speech tagging. An evaluation of sophisticated lexical resources including corpuses and lexicons was carried out, and investigations were conducted on sentiment analysis constructs such as opinion words, modifiers, negations. Results and conclusions: Performance is reported for each of the reviewed study. Based on experimental results and proposals forwarded through this paper provides the groundwork for further studies on Urdu sentiment analysis.

Although over 169 million people in the world are familiar with the Urdu language and a large quantity of Urdu data is being generated on different social websites daily, very few research studies and efforts have been completed to build language resources for the Urdu language and examine user sentiments. The primary objective of this study is twofold: (1) develop a benchmark dataset for resource-deprived Urdu language for sentiment analysis and (2) evaluate various machine and deep learning algorithms for sentiment. To find the best technique, we compare two modes of text representation: count-based, where the text is represented using word $n$ -gram feature vectors and the second one is based on fastText pre-trained word embeddings for Urdu. We consider a set of machine learning classifiers (RF, NB, SVM, AdaBoost, MLP, LR) and deep leaning classifiers (1D-CNN and LSTM) to run the experiments for all the feature types. Our study shows that the combination of word $n$ -gram features with LR outperformed other classifiers for sentiment analysis task, obtaining the highest F1 score of 82.05% using combination of features.

The dawn of the internet opened the doors to the easy and widespread sharing of information on subject matters such as products, services, events and political opinions. While the volume of studies conducted on sentiment analysis is rapidly expanding, these studies mostly address English language concerns. The primary goal of this study is to present state-of-art survey for identifying the progress and shortcomings saddling Urdu sentiment analysis and propose rectifications. We described the advancements made thus far in this area by categorising the studies along three dimensions, namely: text pre-processing lexical resources and sentiment classification. These pre-processing operations include word segmentation, text cleaning, spell checking and part-of-speech tagging. An evaluation of sophisticated lexical resources including corpuses and lexicons was carried out, and investigations were conducted on sentiment analysis constructs such as opinion words, modifiers, negations. Performance is reported for each of the reviewed study. Based on experimental results and proposals forwarded through this paper provides the groundwork for further studies on Urdu sentiment analysis. [ABSTRACT FROM AUTHOR]

From the last decade, Sentiment Analysis of languages such as English and Chinese are particularly the focus of attention but resource poor languages such as Urdu are mostly ignored by the research community, which is focused in this research. After acquiring data from various blogs of about 14 different genres, the data is being annotated with the help of human annotators. Three well-known classifiers, that is, Support Vector Machine, Decision tree and -Nearest Neighbor (-NN) are tested, their outputs are compared and their results are ultimately improved in several iterations after taking a number of steps that include stop words removal, feature extraction, identification and extraction of important features. extraction. Initially, the performance of the classifiers is not satisfactory as the accuracy achieved by all the three is below 50%. Ensemble of classifiers is also tried but the results are not fruitful (in terms of high accuracy). The results are analyzed carefully and improvements are made including feature extraction that raised the performance of these classifiers to a satisfactory level. It is further concluded that -NN is performing better than Support Vector Machine and Decision tree in terms of accuracy, precision, recall and -measure. [ABSTRACT FROM AUTHOR]

Sentiment analysis (SA) can help in decision making, drawing conclusion, or recommending appropriate solution for different business, political, or other problems. At the same time reliable ways are also required to verify the results that are achieved after SA. In the frame of biologically inspired approaches for machine learning, getting reliable result is challenging but important. Properly verified and validated results are always appreciated and preferred by the research community. The strategy of achieving reliable result is adopted in this research by using three standard evaluation measures. First, SA of Urdu is performed. After collection and annotation of data, five classifiers, i.e., PART, Naives Bayes mutinomial Text, Lib SVM (support vector machine), decision tree (J48), and k nearest neighbor (KNN, IBK) are employed using Weka. After using 10-fold cross-validation, three top most classifiers, i.e., Lib SVM, J48, and IBK are selected on the basis of high accuracy, precision, recall, and F-measure. Further, IBK resulted as the best classifier among the three. For verification of this result, labels of the sentences (positive, negative, or neutral) are predicted by using training and test data, followed by the application of the three standard evaluation measures, i.e., McNemar's test, kappa statistic, and root mean squared error. IBK performs much better than the other two classifiers. To make this result more reliable, a number of steps are taken including the use of three evaluation measures for getting a confirmed and validated result which is the main contribution of this research. It is concluded with confidence that IBK is the best classifier in this case. [ABSTRACT FROM AUTHOR]

Just like other languages, a large number of intensifiers are used in Urdu language. There may be a single occurrence of an intensifier or there may be multiple consecutive occurrences. While performing sentiment analysis of Urdu text, these intensifiers need special treatment for obtaining more accurate results, which is the main focus of this research work. A wide coverage Urdu sentiment lexicon is developed where intensifiers are identified and placed in a separate file. While developing Urdu sentiment analyser, rules are specifically formulated for assigning polarities to the intensifiers in text, if they are surrounded by the positive or negative words. Results show that the method proved to be effective in attaining the correct classification of Urdu sentences as positive, negative, or neutral, compared with traditional methods. Implementation of rules for intensifiers increased the accuracy of Urdu sentiment analyser from 78.33% to 83.42%, which is a statistically significant improvement. It is concluded that intensifiers cannot be ignored while performing sentiment analysis. Effective handling of intensifiers can significantly improve the performance of sentiment analyser. [ABSTRACT FROM AUTHOR]