Review of the Phonological System of Contemporary Urdu Spoken in Pakistan.

PURPOSE: Urdu is the lingua franca and national language of Pakistan, and is the 10th most-spoken language worldwide with over 230 million speakers. The Urdu phonological system has been examined over the past decades. However, the system has been evolving. This paper aimed to review the available studies investigating various aspects of the Urdu phonological system and to reveal the variations noted among these studies. METHOD: Twenty-one studies examining the phonological system of Urdu were located. The studies were reviewed in terms of consonants, geminates, consonant clusters, vowels, diphthongs, syllable structure, phonotactic constraints, and stress. RESULT: The findings indicated that 38 consonants, 23 vowels, and 15 diphthongs are used in contemporary Urdu. Most consonants exist as geminates word medially. There are six syllable structures. The consonant clusters are constrained to the coda position only, and short vowels cannot exist in the word-final position. Like other syllable-timed languages, stress is not prominent in Urdu. CONCLUSION: Based on this review, a contemporary Urdu phonemic and syllable structure inventory has been proposed. This will serve as a reference for use in further acquisition research and clinical practice.

Assessment of physical literacy in 8- to 12-year-old Pakistani school children: reliability and cross-validation of the Canadian assessment of physical literacy-2 (CAPL-2) in South Punjab, Pakistan.

BACKGROUND: The increasing prevalence of physical inactivity, declining fitness, and rising childhood obesity highlight the importance of physical literacy (PL), as a foundational component for fostering lifelong health and active lifestyle. This recognition necessitates the development of effective tools for PL assessment that are applicable across diverse cultural landscapes. AIM: This study aimed to translate the Canadian Assessment of Physical Literacy-2 (CAPL-2) into Urdu and adapt it for the Pakistani cultural context, to assess PL among children aged 8-12 years in Pakistan. METHOD: The Urdu version of CAPL-2 was administered among 1,360 children aged 8-12 from 87 higher secondary schools across three divisions in South Punjab province, Pakistan. Statistical analysis includes test-retest reliability and construct validity, employing confirmatory factor analysis to evaluate the tool's performance both overall and within specific subdomains. RESULTS: The Urdu version of CAPL-2 demonstrated strong content validity, with a Content Validity Ratio of 0.89. Confirmatory factor analysis supported the four-factor structure proposed by the original developers, evidenced by excellent model fit indices (GFI = 0.984, CFI = 0.979, TLI = 0.969, RMSEA = 0.041). High internal consistency was observed across all domains (α = 0.988 to 0.995), with significant correlations among most, excluding the Knowledge and Understanding domains. Notably, gender and age significantly influenced performance, with boys generally scoring higher than girls, with few exceptions. CONCLUSION: This study marks a significant step in the cross-cultural adaptation of PL assessment tools, successfully validating the CAPL-2 Urdu version for the Pakistani context for the first time. The findings affirm the tool's suitability for assessing PL among Pakistani children, evidencing its validity and reliability across the Pakistani population.

Identifying developmental language disorder (DLD) in multilingual children: A case study tutorial.

PURPOSE: A long-standing issue in identifying developmental language disorder (DLD) in multilingual children is differentiating between effects of language experience and genuine impairment when clinicians often lack suitable norm-referenced assessments. In this tutorial we demonstrate, via a case study, that it is feasible to identify DLD in a multilingual child using the CATALISE diagnostic criteria, Language Impairment Testing in Multilingual Settings (LITMUS) assessment tools, and telepractice. METHOD: This tutorial features a case study of one 6-year-old Urdu-Cantonese multilingual ethnic minority child, and seven age- and grade-matched multilinguals. They were tested via Zoom using Urdu versions of the Multilingual Assessment Instrument for Narratives (LITMUS-MAIN), the Crosslinguistic Lexical Task (LITMUS-CLT), the Crosslinguistic Nonword Repetition Test (LITMUS-CL-NWR), and the Sentence Repetition Task (LITMUS-SRep). RESULT: The child scored significantly lower in the LITMUS tests compared to her peers in her best/first language of Urdu. Together with the presence of negative functional impact and poor prognostic features, and absence of associated biomedical conditions, the findings suggest this participant could be identified as having DLD using the CATALISE diagnostic criteria. CONCLUSION: The result demonstrates the promise of this approach to collect reference data and identify DLD in multilingual children. The online LITMUS battery has the potential to support identification of multilingual DLD in any target language.

Addressing cyberbullying in Urdu tweets: a comprehensive dataset and detection system.

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.

Adapting multilingual vision language transformers for low-resource Urdu optical character recognition (OCR).

In the realm of digitizing written content, the challenges posed by low-resource languages are noteworthy. These languages, often lacking in comprehensive linguistic resources, require specialized attention to develop robust systems for accurate optical character recognition (OCR). This article addresses the significance of focusing on such languages and introduces ViLanOCR, an innovative bilingual OCR system tailored for Urdu and English. Unlike existing systems, which struggle with the intricacies of low-resource languages, ViLanOCR leverages advanced multilingual transformer-based language models to achieve superior performances. The proposed approach is evaluated using the character error rate (CER) metric and achieves state-of-the-art results on the Urdu UHWR dataset, with a CER of 1.1%. The experimental results demonstrate the effectiveness of the proposed approach, surpassing state of the-art baselines in Urdu handwriting digitization.

The Translation and Preliminary Psychometric Validation of the Ghosting Questionnaire in Urdu.

BACKGROUND: "Ghosting" refers to the practice of abruptly cutting off all contact with a person with whom you have been in constant correspondence. The break comes without warning and without understandable provocation. The term most commonly applies to online romantic relationships. The motives for and effects of ghosting have been studied, and validated research questionnaires have been developed; however, there are no such questionnaires available for Urdu speakers. The purpose of this study was to adapt the "Ghosting Questionnaire (GQ)" for use in Pakistan and India, two of the world's most populous countries-a process that involves translation, adaptation, and validation. METHODS: The study's methodology involved translating the GQ into Urdu using both forward and backward translation techniques. Convergent validity, test-retest reliability, internal consistency, confirmatory factor analysis, and goodness of fit were all components of the psychometric analyses. CONCLUSIONS: The Urdu version of the GQ demonstrated a good internal consistency, with the Cronbach's alpha and McDonald's omega both exceeding 0.90. It also showed a high test-retest reliability-(0.96). The one-factor structure was confirmed by the confirmatory factor analysis, which agreed with the original English version of the GQ.

Adaptation and validation of the west haven-yale multidimensional pain inventory for Pakistani population.

Objective: To evaluate the adapted version of West Haven-Yale Multidimensional Pain Inventory for patients with chronic pain. METHODS: The multiphase study was conducted from January to December 2021. The factorial structure of the Urdu version of West Haven-Yale Multidimensional Pain Inventory was evaluated on a sample of adult patients aged 18- 45 years with non-specific chronic pain, taken from public and private hospitals and clinics of Lahore, Pakistan. The Urdu version was then subjected to factor analysis, while Cronbach's alpha, composite reliability, convergent and discriminant validity of the scale were also calculated. Data was analysed using SPSS 24. RESULTS: Of the 306 subjects, 204(66.7%) were females and 102(33.3%) were men. The overall mean age was 30.94+/-8.44 years. There were 166(54.2%) subjects who were married, and 137(44.8%) reported experiencing pain daily. The confirmatory factor analysis showed a 45-item structure for 12 sub-scales as the best fit. The statistics for the final model were observed as minimum discrepancy function by degrees of freedom divided was 1.69, root mean square error of approximation was 0.05, and standardised root mean square residual was 0.06. Comparative fit index value was 0.91 and Tucker-Lewis coefficient was 0.90. Cronbach's alpha reliability ranged between 0.68 and 0.89 for the subscales, while for the total scale, it was 0.72. Conclusion: The Urdu version of West Haven-Yale Multidimensional Pain Inventory was found to be a reliable and valid tool for chronic pain assessment for patients in Pakistan.

A diachronic study determining syntactic and semantic features of Urdu-English neural machine translation.

Machine translation produces marginal accuracy rates for low-resource languages, but its deep learning model expects to yield improved accuracy with time. This longitudinal study investigates how Google Translate's Urdu-to-English translated output has evolved between 2018 and 2021. Accuracy and acceptability of the translations have been determined by, a) an interlinear gloss that identifies core semantic units and grammatical functions to be translated and, b) a descriptive comparison of the translated text's syntactic and semantic properties with those of the source text. Overall, despite a 50 % error rate that persists over the three-year interval, the research reports significant improvement in the overall intelligibility of the translations, in contrast to initial results from 2018, which exhibited rampant non-localized errors. Working backwards from instances of errors to morphosyntactic and semantic patterns underlying them, the study concludes that the pro-drop feature of Urdu, Urdu's case-marking system, identification of clause boundaries, polysemous terms, and orthographically similar words pose the greatest difficulty in neural machine translation. These results point to the need for incorporating syntactic information in training data.

A versatile dataset for intrinsic plagiarism detection, text reuse analysis, and author clustering in Urdu.

Plagiarism detection (PD) is a process of identifying instances where someone has presented another person's work or ideas as their own. Plagiarism detection is categorized into two types (i) Intrinsic plagiarism detection primarily concerns the assessment of authorship consistency within a single document, aiming to identify instances where portions of the text may have been copied or paraphrased from elsewhere within the same document. Author clustering, closely related to intrinsic plagiarism detection, involves grouping documents based on their stylistic and linguistic characteristics to identify common authors or sources within a given dataset. On the other hand, (ii) extrinsic plagiarism detection delves into the comparative analysis of a suspicious document against a set of external source documents, seeking instances of shared phrases, sentences, or paragraphs between them, which is often referred to as text reuse or verbatim copying. Detection of plagiarism from documents is a long-established task in the area of NLP with remarkable contributions in multiple applications. A lot of research has already been conducted in the English and other foreign languages but Urdu language needs a lot of attention especially in intrinsic plagiarism detection domain. The major reason is that Urdu is a low resource language and unfortunately there is no high-quality benchmark corpus available for intrinsic plagiarism detection in Urdu language. This study presents a high-quality benchmark Corpus comprising 10,872 documents. The corpus is structured into two granularity levels: sentence level and paragraph level. This dataset serves multifaceted purposes, facilitating intrinsic plagiarism detection, verbatim text reuse identification, and author clustering in the Urdu language. Also, it holds significance for natural language processing researchers and practitioners as it facilitates the development of specialized plagiarism detection models tailored to the Urdu language. These models can play a vital role in education and publishing by improving the accuracy of plagiarism detection, effectively addressing a gap and enhancing the overall ability to identify copied content in Urdu writing.

Normalized effect size (NES): a novel feature selection model for Urdu fake news classification.

Social media has become an essential source of news for everyday users. However, the rise of fake news on social media has made it more difficult for users to trust the information on these platforms. Most research studies focus on fake news detection in the English language, and only a limited number of studies deal with fake news in resource-poor languages such as Urdu. This article proposes a globally weighted term selection approach named normalized effect size (NES) to select highly discriminative features for Urdu fake news classification. The proposed model is based on the traditional inverse document frequency (TF-IDF) weighting measure. TF-IDF transforms the textual data into a weighted term-document matrix and is usually prone to the curse of dimensionality. Our novel statistical model filters the most discriminative terms to reduce the data's dimensionality and improve classification accuracy. We compare the proposed approach with the seven well-known feature selection and ranking techniques, namely normalized difference measure (NDM), bi-normal separation (BNS), odds ratio (OR), GINI, distinguished feature selector (DFS), information gain (IG), and Chi square (Chi). Our ensemble-based approach achieves high performance on two benchmark datasets, BET and UFN, achieving an accuracy of 88% and 90%, respectively.

Reliability and validity of Urdu PARENTS for assessing non-technical skills of paediatric residents in a teaching hospital in Pakistan.

PURPOSE: The primary objective of our study is twofold. First, we assessed nontechnical skills (NTSs), such as the cognitive, social, and personal skills of postgraduate residents (PGRs), from paediatric caregivers' perspectives in a paediatric emergency department (PED) of a tertiary care hospital. Second, we evaluated the reliability and validity of the 'Parents' Assessment of Residents Enacting Non-Technical Skills' (PARENTS) instrument in its Urdu-translated version, ensuring its applicability and accuracy in the Pakistani context. MATERIALS AND METHODS: This mixed-method study used an instrument translation and validation design. We translated an existing instrument, PARENTS, into Urdu, the national language of Pakistan, and administered it to paediatric caregivers in the PED of a tertiary care hospital. We collected data from 471 paediatric caregivers and coded them for analysis in AMOS and SPSS. RESULTS: The Urdu-translated version of the PARENTS demonstrated reliability and internal validity in our study. The findings from the assessment revealed that paediatric caregivers expressed satisfaction with the knowledge and skill of residents. However, there was comparatively lower satisfaction regarding the residents' display of patience or empathy towards the children under their care. CONCLUSION: The study findings support the validity and reliability of the PARENTS as an effective instrument for assessing the NTS of PGRs from the perspective of paediatric caregivers. With its demonstrated efficacy, medical educators can utilize PARENTS to pinpoint specific areas that require attention regarding the NTS of PGRs, thus facilitating targeted interventions for enhanced patient care outcomes.

Psychometric evaluation of Urdu-translated Amsterdam preoperative anxiety and information scale and assessment of preoperative anxiety in adult surgical patients of Karachi, Pakistan: A cross-sectional study.

Objective: The primary aim of this study was to develop an Urdu-translated version of the Amsterdam preoperative anxiety and information scale and perform a psychometric evaluation of it. A secondary aim was to estimate the prevalence of preoperative anxiety using Urdu-translated Amsterdam preoperative anxiety and information scale in patients undergoing surgery in Karachi, Pakistan, and the factors contributing to anxiety among them. Method: This cross-sectional survey included 267 patients enrolled for elective surgery under general anesthesia from March 5 to November 20, 2022. In psychometric analysis, face validity, criterion validity, construct validity, and reliability of Urdu-translated Amsterdam preoperative anxiety and information scale were determined. Face validity was evaluated by performing blind-back translation and a pilot study. Criterion validity was evaluated by correlating the Amsterdam preoperative anxiety and information scale with the visual analog scale for anxiety. Exploratory factor analysis and Cronbach's α test were used to analyze construct validity and reliability, respectively. The associate variables were identified by performing a one-sample t-test and one-way analysis of variance on SPSS 26. Results: Cronbach's α test is 0.85 for the Amsterdam preoperative anxiety and information scale anxiety scale and 0.70 for the need for information. 65.3% of the total variance is explained by the Urdu version of Amsterdam preoperative anxiety and information scale items in factor analysis and the intercorrelation of all items was >0.20 (mean: 0.575). Urdu-translated Amsterdam preoperative anxiety and information scale and visual analog scale for anxiety showed a good correlation (r = 0.664, p < 0.001). The overall prevalence of preoperative anxiety among patients is 52.4% suggested by the Amsterdam preoperative anxiety and information scale cutoff score of more than 11. Females, students, and patients elected for major surgery shared significantly higher anxiety levels (p < 0.05). The commonest factors contributing to anxiety are postoperative pain in 140 (52.4%) patients, fear of death in 115 (43.1%), and financial loss in 91 (34.1%). Conclusions: The Urdu-translated Amsterdam preoperative anxiety and information scale is a reliable, valid, and acceptable screening tool for preoperative anxiety. The prevalence of preoperative anxiety was high. The preoperative anxiety level is significantly associated with gender, employment status, and type of surgery.

Translation and cross-cultural adaptation of EAT-26 questionnaire in Urdu.

EAT-26 questionnaire is used globally to screen people for the risk of eating disorders. In addition to this, the EAT-26 is being used in its original English version in research and clinical settings due to unavailability of the Urdu version. Therefore, the aim of this study was to introduce the Urdu version of EAT-26 to clinicians and academicians in Pakistan, interested in the assessment of population at risk of eating disorders. After getting the formal permission for translation by Dr. D. M. Garner, WHO guidelines were followed for the translation and adaptation process. Two independent translators with psychological background worked under the supervision of a lead to produce the definitive version following six steps of translation and adaptation. Cognitive interviews and focused group discussions helped in the assessment process for the understanding level of translated Urdu version. The pre-final version showed comprehension and acceptability during initial pilot testing.•The final translated version of EAT-26 in Urdu will be available on Internet to use. It is expected that the use of EAT-26 will be widespread in Pakistan, aiming at the assessment of eating disorders.•The Urdu version of EAT-26 is finalized, and ready to use by researchers and clinicians in Pakistan.

The translation process of the culturally sensitive active aging scale for community-dwelling older adults in Pakistan.

BACKGROUND: The assessment of active aging levels in Pakistani older adults is crucial yet; research tools are scarce in the local language. Therefore, this study aims to translate and validate the English version of an Active Aging Scale into a cross-culturally sensitive Urdu version to assess active aging levels in Pakistani older adults. METHODS: To translate and validate the scale, we used the ISPOR (International Society for Pharmacy Economic and Outcome Research) standards. Reliability, concurrent validity, construct validity, convergent validity, and discriminatory validity were checked on a total sample of 160 community-dwelling older adults. After two weeks, the test-retest reliability was examined. AMOS version 23 and SPSS version 23 were used to analyze the data. RESULTS: The average content validity index for clarity was 0.91 and relevancy was 0.80. The total variance in the pilot study of all items secured > 0.3 variances except for two items scored < 0.30 that were omitted before the validity and reliability test. The remaining items explained 65.46% of the overall variation and had factor loadings ranging from 0.46 to 0.90 in the principal factor analysis (PFA). The confirmatory factor analysis of the Active Aging Scale revealed that the model fit was good with a Chi-square value (418.18 (DF = 2.2) which is less than 3.00. This is further evidenced by the root mean square error of approximation (RMSEA) of 0.042, goodness of fit index (GFI) of 0.92, adjusted goodness of fit index (AGFI) of 0.94, and comparative fit index (CFI) values of 0.92 and 0.96 (unstandardized and standardized, respectively). The scale's Cronbach's alpha coefficient was 0.88, indicating dependability and its test-retest reliability with the significance of (P. < 0.05). CONCLUSION: The Urdu version of the Active Aging Scale was successfully translated and validated in a culturally sensitive manner, and can be used to evaluate the effectiveness of various active aging interventions for older adults in Pakistan.

Translation and validation of urdu version short form-mcgill pain questionnaire-2.

BACKGROUND: Low back pain is one of the most common complaints affecting many individuals. The McGill Pain Questionnaire is used in various clinical settings to assess different types of pain and one of the most extensively used outcomes measures for pain in the world. The purpose of this study was to translate and validate the original English version of the SF-MPQ-2 into Urdu (SF-MPQ-2-U). METHODS: For this study, Mapi Research Trust protocols were followed for the forward and backward translation. Test-retest reliability was used to assess the reliability. Cronbach's alpha and Omega was used to determine internal consistency. Pearson's correlation was used to evaluate convergent validity. Confirmatory factor analysis was also conducted. RESULTS: The Cronbach's alpha for SF-MPQ-2-U was 0.73 to 0.79, indicating acceptable internal consistency. Omega score for the SF-MPQ-U were 0.918. The ICC varied from 0.799 to 0.878 for domains of SF-MPQ-2-U. The CFA of the SF-MPQ-2-U met model fit indices with GFI and NFI > 0.90. The inter-scale correlation between baseline and re-test data was from 0.63 to 0.71, indicating a positive and strong correlation. The SF-MPQ-2-U and ODI-U had a baseline correlation of 0.547. The correlation of SF-MPQ-2-U & VAS at baseline data was 0.558. Pearson's correlation between subscales was r = 0.253 with p 0.01, which was statistically significant. CONCLUSION: The SF-MPQ-2-U is considered to have good convergent validity at inter scale and between two scale levels. Reliability was checked by test-retest reliability, Internal consistency was checked using Cronbach's alpha and Omega that showed good internal consistency for measuring different types of pain in patients with low back pain who speak Urdu. To make the questionnaire more valid and reliable, it is recommended for the researchers to do in-depth research on larger sample size.

Fake news detection in Urdu language using machine learning.

With the rise of social media, the dissemination of forged content and news has been on the rise. Consequently, fake news detection has emerged as an important research problem. Several approaches have been presented to discriminate fake news from real news, however, such approaches lack robustness for multi-domain datasets, especially within the context of Urdu news. In addition, some studies use machine-translated datasets using English to Urdu Google translator and manual verification is not carried out. This limits the wide use of such approaches for real-world applications. This study investigates these issues and proposes fake news classier for Urdu news. The dataset has been collected covering nine different domains and constitutes 4097 news. Experiments are performed using the term frequency-inverse document frequency (TF-IDF) and a bag of words (BoW) with the combination of n-grams. The major contribution of this study is the use of feature stacking, where feature vectors of preprocessed text and verbs extracted from the preprocessed text are combined. Support vector machine, k-nearest neighbor, and ensemble models like random forest (RF) and extra tree (ET) were used for bagging while stacking was applied with ET and RF as base learners with logistic regression as the meta learner. To check the robustness of models, fivefold and independent set testing were employed. Experimental results indicate that stacking achieves 93.39%, 88.96%, 96.33%, 86.2%, and 93.17% scores for accuracy, specificity, sensitivity, MCC, ROC, and F1 score, respectively.

Abstractive text summarization of low-resourced languages using deep learning.

Background: Humans must be able to cope with the huge amounts of information produced by the information technology revolution. As a result, automatic text summarization is being employed in a range of industries to assist individuals in identifying the most important information. For text summarization, two approaches are mainly considered: text summarization by the extractive and abstractive methods. The extractive summarisation approach selects chunks of sentences like source documents, while the abstractive approach can generate a summary based on mined keywords. For low-resourced languages, e.g., Urdu, extractive summarization uses various models and algorithms. However, the study of abstractive summarization in Urdu is still a challenging task. Because there are so many literary works in Urdu, producing abstractive summaries demands extensive research. Methodology: This article proposed a deep learning model for the Urdu language by using the Urdu 1 Million news dataset and compared its performance with the two widely used methods based on machine learning, such as support vector machine (SVM) and logistic regression (LR). The results show that the suggested deep learning model performs better than the other two approaches. The summaries produced by extractive summaries are processed using the encoder-decoder paradigm to create an abstractive summary. Results: With the help of Urdu language specialists, the system-generated summaries were validated, showing the proposed model's improvement and accuracy.

Translation and validation of the Pelvic Girdle Questionnaire in the Urdu language.

INTRODUCTION AND HYPOTHESIS: The Pelvic Girdle Questionnaire (PGQ) is designed to determine pain and limited activities in pregnant and postpartum women with pelvic girdle pain (PGP). The purpose of this study was to translate the PGQ into the Urdu language and find out the cross-cultural validity and reliability of the Urdu PGQ among pregnant and postpartum women. METHODS: A translation and cultural adaptation study was performed following international guidelines. The PGQ (the Urdu version) was drafted and a pilot study was conducted on 16 pregnant and postpartum women. A total of 125 pregnant and postpartum females participated in this study for validity and test-retest reliability. SPSS 25 was used for data analysis. RESULTS: Content validity was analyzed by the content validity index ranging from (0.92 to 1). Convergent validity was determined by correlating the Urdu version of the PGQ with the Oswestry Disability Index (ODI). Pearson rank correlation coefficient between the PGQ and the ODI (p=0.84) showed convergent validity. Internal consistency and test-retest reliability of the Urdu version of the PGQ were calculated by Cronbach's alpha (α=0.98), and intra-class correlation coefficient (ICC=0.98) respectively. CONCLUSION: The Urdu version of the PGQ showed good content and convergent validity as well as high internal consistency and test-retest reliability.

Roman Urdu Hate Speech Detection Using Transformer-Based Model for Cyber Security Applications.

Social media applications, such as Twitter and Facebook, allow users to communicate and share their thoughts, status updates, opinions, photographs, and videos around the globe. Unfortunately, some people utilize these platforms to disseminate hate speech and abusive language. The growth of hate speech may result in hate crimes, cyber violence, and substantial harm to cyberspace, physical security, and social safety. As a result, hate speech detection is a critical issue for both cyberspace and physical society, necessitating the development of a robust application capable of detecting and combating it in real-time. Hate speech detection is a context-dependent problem that requires context-aware mechanisms for resolution. In this study, we employed a transformer-based model for Roman Urdu hate speech classification due to its ability to capture the text context. In addition, we developed the first Roman Urdu pre-trained BERT model, which we named BERT-RU. For this purpose, we exploited the capabilities of BERT by training it from scratch on the largest Roman Urdu dataset consisting of 173,714 text messages. Traditional and deep learning models were used as baseline models, including LSTM, BiLSTM, BiLSTM + Attention Layer, and CNN. We also investigated the concept of transfer learning by using pre-trained BERT embeddings in conjunction with deep learning models. The performance of each model was evaluated in terms of accuracy, precision, recall, and F-measure. The generalization of each model was evaluated on a cross-domain dataset. The experimental results revealed that the transformer-based model, when directly applied to the classification task of the Roman Urdu hate speech, outperformed traditional machine learning, deep learning models, and pre-trained transformer-based models in terms of accuracy, precision, recall, and F-measure, with scores of 96.70%, 97.25%, 96.74%, and 97.89%, respectively. In addition, the transformer-based model exhibited superior generalization on a cross-domain dataset.

Translation and validation of Incontinence Impact Questionnaire Short Form in the Urdu language.

INTRODUCTION AND HYPOTHESIS: The Incontinence Impact Questionnaire Short Form (IIQ-7 SF) is a self-administered questionnaire that is used to assess the impact of urinary incontinence on the quality of life in women. It is translated into different languages; however, there is currently no official Urdu version of this tool. The main purpose of this study was to translate the IIQ-7 SF into the Urdu language and to find out its validity and reliability in women with urinary incontinence. METHODS: The IIQ-7 was translated into the Urdu language by following the standardized steps. The original version was translated into Urdu by two translators and the back translation into English was done by an independent translator. A panel of experts reviewed the translations and a final version was drafted. Fifteen women with urinary incontinence were involved in the pilot study. The validity and reliability were then assessed on 70 women with urinary incontinence. RESULTS: The content validity index (CVI) of each question ranged from 0.91 to 0.94. The convergent validity with UDI-6 was determined by using Spearman's correlation coefficient (r=0.90). Cronbach's α showed the internal consistency, which is 0.87. The test-retest reliability was calculated by the intra-class correlation coefficient, ICC=0.95. The scree plot showed the two components have eigen values greater than 1. CONCLUSIONS: The Urdu version of the IIQ-7 has shown good validity and reliability in incontinence patients, according to the findings.