K. Nandhini
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
K. Nandhini has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
- Bharathi Bhagavathsingh
Department of Computer Science and Engineering, Sri Sivasubramaniya Nadar College of Engineering, Chennai, India
2 shared publications - Charmathi Rajkumar
The American College, Madurai, India
2 shared publications - Durairaj Thenmozhi
Sri Sivasubramaniya Nadar College of Engineering
2 shared publications - Kishore Kumar Ponnusamy
Digital University of Kerala, Thiruvananthapuram, India
2 shared publications - N. Sripriya
Sri Sivasubramaniya Nadar College of Engineering
2 shared publications - P. Senthilkumar
Department of Computer Science, Central University of Tamil Nadu, Tamil Nadu, India
2 shared publications
Latest Publications
- Open Access
- 2 October 2026
- Published by Engineering, Technology & Applied Science Research in Engineering, Technology & Applied Science Research
- Vol. 16 (5) , 39649-39656
A Context-Aware Subjective Question Generation Model Using a Domain-Specific Fine-Tuned T5 Transformer
Mock tests play an important role in preparing for academic and competitive examinations, and most mock tests have been designed based on the Question Answering (QA) technique, which is one of the areas in Natural Language Processing (NLP). There are two main tasks in QA: Automatic Question Generation (AQG), which entails generating questions from a given context based on available keywords, and Automatic Answer Generation (AAG), which entails generating an answer based on the given question and context. To build a QA system, a proper dataset is required, and a domain-specific dataset can provide advantages over a generic-domain dataset in terms of processing time, hardware requirements, and accuracy. However, fewer domain-specific datasets are available for subjective computer science examinations. To fulfill this need, we develop a question generation model for building a domain-specific dataset. The proposed approach consists of four phases. In the first phase, a hybrid training dataset is constructed by curating subjective questions from two established benchmarks, the Stanford Question Answering Dataset (SQuAD) and the ReAding Comprehension Dataset from Examinations (RACE). In the second phase, computer science-related data are collected from Wikipedia, comprising 406 computer science topics with 17,520 sentences. In the third phase, the Text-to-Text Transfer Transformer (T5) model is utilized with the hybrid dataset to generate subjective questions. In the fourth phase, the generated questions are evaluated using Bilingual Evaluation Understudy (BLEU), Recall-Oriented Understudy for Gisting Evaluation (ROUGE-L), and Metric for Evaluation of Translation with Explicit ORdering (METEOR), as well as human judgment based on a five-point Likert scale to assess the quality of the generated questions and the consistency of the human evaluations. The proposed model generates subjective questions for the computer science dataset and outperforms other baseline models according to BLEU, ROUGE-L, and METEOR scores, as well as human annotator scores.
- 2 October 2026
- Published by Engineering, Technology & Applied Science Research in Engineering, Technology & Applied Science Research
- Vol. 16 (5) , 39649-39656
Open AccessA Context-Aware Subjective Question Generation Model Using a Domain-Specific Fine-Tuned T5 Transformer
Mock tests play an important role in preparing for academic and competitive examinations, and most mock tests have been designed based on the Question Answering (QA) technique, which is one of the areas in Natural Language Processing (NLP). There are two main tasks in QA: Automatic Question Generation (AQG), which entails generating questions from a given context based on available keywords, and Automatic Answer Generation (AAG), which entails generating an answer based on the given question and context. To build a QA system, a proper dataset is required, and a domain-specific dataset can provide advantages over a generic-domain dataset in terms of processing time, hardware requirements, and accuracy. However, fewer domain-specific datasets are available for subjective computer science examinations. To fulfill this need, we develop a question generation model for building a domain-specific dataset. The proposed approach consists of four phases. In the first phase, a hybrid training dataset is constructed by curating subjective questions from two established benchmarks, the Stanford Question Answering Dataset (SQuAD) and the ReAding Comprehension Dataset from Examinations (RACE). In the second phase, computer science-related data are collected from Wikipedia, comprising 406 computer science topics with 17,520 sentences. In the third phase, the Text-to-Text Transfer Transformer (T5) model is utilized with the hybrid dataset to generate subjective questions. In the fourth phase, the generated questions are evaluated using Bilingual Evaluation Understudy (BLEU), Recall-Oriented Understudy for Gisting Evaluation (ROUGE-L), and Metric for Evaluation of Translation with Explicit ORdering (METEOR), as well as human judgment based on a five-point Likert scale to assess the quality of the generated questions and the consistency of the human evaluations. The proposed model generates subjective questions for the computer science dataset and outperforms other baseline models according to BLEU, ROUGE-L, and METEOR scores, as well as human annotator scores.
- Open Access
- 5 December 2025
- Published by Centre for Research and Innovation in Asian Journal of Computer Science and Technology
- Vol. 14 (2) , 51-55
N-gram Based Lexical Sentence Similarity Score Using Modified Jaccard Algorithm
In recent years, the Internet has evolved into a global phenomenon, making it nearly impossible to envision modern life without it. Among the vast forms of online content, textual data holds the greatest significance due to its abundance and informational value. However, managing and analysing such extensive text corpora poses several challenges, with sentence similarity emerging as one of the most complex problems in Natural Language Processing (NLP). Although existing sentence comparison techniques perform effectively in specific contexts, they often struggle in others and typically require substantial computational resources, including powerful hardware, extensive training datasets, and high processing capabilities. To address these limitations, this study introduces a lightweight approach that emphasizes word-level similarity through comprehensive n-gram comparisons. The proposed method incorporates semantic understanding and evaluates the longest common subsequence within sentences to generate a more accurate similarity score. It demonstrates superior efficiency over baseline methods by minimizing computational requirements and leveraging straightforward mathematical operations.
- 5 December 2025
- Published by Centre for Research and Innovation in Asian Journal of Computer Science and Technology
- Vol. 14 (2) , 51-55
Open AccessN-gram Based Lexical Sentence Similarity Score Using Modified Jaccard Algorithm
In recent years, the Internet has evolved into a global phenomenon, making it nearly impossible to envision modern life without it. Among the vast forms of online content, textual data holds the greatest significance due to its abundance and informational value. However, managing and analysing such extensive text corpora poses several challenges, with sentence similarity emerging as one of the most complex problems in Natural Language Processing (NLP). Although existing sentence comparison techniques perform effectively in specific contexts, they often struggle in others and typically require substantial computational resources, including powerful hardware, extensive training datasets, and high processing capabilities. To address these limitations, this study introduces a lightweight approach that emphasizes word-level similarity through comprehensive n-gram comparisons. The proposed method incorporates semantic understanding and evaluates the longest common subsequence within sentences to generate a more accurate similarity score. It demonstrates superior efficiency over baseline methods by minimizing computational requirements and leveraging straightforward mathematical operations.
- 1 September 2025
- Published by Springer Nature
- p. 365-375
A Fuzzy-Based Subjective Answer Assessment Grading System Using Sentence Transformer
- 1 September 2025
- Published by Springer Nature
- p. 365-375
A Fuzzy-Based Subjective Answer Assessment Grading System Using Sentence Transformer
- Open Access
- 12 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 3-6
Insights into Sarcasm Detection in Social Media Posts for Dravidian Languages: Tamil and Malayalam
- 12 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 3-6
Open AccessInsights into Sarcasm Detection in Social Media Posts for Dravidian Languages: Tamil and Malayalam
- 24 April 2024
- Published by Springer Nature
- p. 236-245
Telugu-English Abusive Comment Detection Using XLMRoBERTa and mBERT
- 24 April 2024
- Published by Springer Nature
- p. 236-245
Telugu-English Abusive Comment Detection Using XLMRoBERTa and mBERT
- Open Access
- 25 March 2024
- Published by Dr. M.G.R. Educational and Research Institute in International Journal of Engineering Research and Sustainable Technologies (IJERST)
- Vol. 2 (1) , 29-35
MITIGATING CLASS IMBALANCE IN OFFENSIVE LANGUAGE DETECTION IN MALAYALAM THROUGH NLPAUG
The rise of technology alongside the prevalence of social media and the promotion of free speech has resulted in an increased presence of vulnerable content in the public sphere. Currently, various researches demonstrate that the identification of offensive language plays a crucial role in preventing or protecting vulnerable groups. Our attention is directed towards the detection of offensive language in Malayalam, recognizing the scarcity of existing research in this area for the Malayalam language. mBERT demonstrates effectiveness across Indian languages. To address class imbalances within datasets, we employed NlpAug for word-level augmentation and achieved a significant improvement in macro F1 score of 0.31.
- 25 March 2024
- Published by Dr. M.G.R. Educational and Research Institute in International Journal of Engineering Research and Sustainable Technologies (IJERST)
- Vol. 2 (1) , 29-35
Open AccessMITIGATING CLASS IMBALANCE IN OFFENSIVE LANGUAGE DETECTION IN MALAYALAM THROUGH NLPAUG
The rise of technology alongside the prevalence of social media and the promotion of free speech has resulted in an increased presence of vulnerable content in the public sphere. Currently, various researches demonstrate that the identification of offensive language plays a crucial role in preventing or protecting vulnerable groups. Our attention is directed towards the detection of offensive language in Malayalam, recognizing the scarcity of existing research in this area for the Malayalam language. mBERT demonstrates effectiveness across Indian languages. To address class imbalances within datasets, we employed NlpAug for word-level augmentation and achieved a significant improvement in macro F1 score of 0.31.
- Open Access
- 15 December 2023
- Published by Association for Computing Machinery (ACM)
- p. 22-24
Findings of Shared task on Sarcasm Identification in Code-Mixed Dravidian Languages
Identifying sarcasm in code-mixed languages presents a pivotal challenge within the domain of natural language processing, given the widespread use of multilingual and multicultural communication on social media platforms. This paper presents findings of the Sarcasm Identification shared task hosted as part of DravidianCodeMix@FIRE-2023. The principal objective of this task was to detect instances of sarcasm within a dataset of code-mixed comments in Tamil-English and Malayalam-English, collected from social media platforms. A total of 11 teams participated in this collaborative effort. This analysis provides a comprehensive assessment of the diverse models employed by the participating teams and the system’s performance was measured in terms of the macro-F1 score.
- 15 December 2023
- Published by Association for Computing Machinery (ACM)
- p. 22-24
Open AccessFindings of Shared task on Sarcasm Identification in Code-Mixed Dravidian Languages
Identifying sarcasm in code-mixed languages presents a pivotal challenge within the domain of natural language processing, given the widespread use of multilingual and multicultural communication on social media platforms. This paper presents findings of the Sarcasm Identification shared task hosted as part of DravidianCodeMix@FIRE-2023. The principal objective of this task was to detect instances of sarcasm within a dataset of code-mixed comments in Tamil-English and Malayalam-English, collected from social media platforms. A total of 11 teams participated in this collaborative effort. This analysis provides a comprehensive assessment of the diverse models employed by the participating teams and the system’s performance was measured in terms of the macro-F1 score.
- 9 October 2023
- Published by IGI Global Scientific Publishing
- p. 303-326
Improving Maternal Health by Predicting Various Pregnancy-Related Abnormalities Using Machine Learning Algorithms
Over the past few decades, artificial intelligence has been showing its high relevance and potential in a vast number of applications, particularly in the healthcare domain. Having a healthy pregnancy is one of the best ways to promote a healthy birth. Getting early and regular prenatal care improves the chances of a healthy pregnancy. Complications involved in the individual's pregnancy need to be predicted on time accurately. AI can help clinicians to make decisions by assisting them in decision-making. In this regard, the objective of this chapter is to provide a detailed survey of various pregnancy-related abnormalities; and to explore various machine learning algorithms to classify/predict pregnancy-related abnormalities with higher accuracy. A generic framework that focuses more on classifying various features into normal and abnormal, and to be monitored patients to provide support and care during an emergency.
- 9 October 2023
- Published by IGI Global Scientific Publishing
- p. 303-326
Improving Maternal Health by Predicting Various Pregnancy-Related Abnormalities Using Machine Learning Algorithms
Over the past few decades, artificial intelligence has been showing its high relevance and potential in a vast number of applications, particularly in the healthcare domain. Having a healthy pregnancy is one of the best ways to promote a healthy birth. Getting early and regular prenatal care improves the chances of a healthy pregnancy. Complications involved in the individual's pregnancy need to be predicted on time accurately. AI can help clinicians to make decisions by assisting them in decision-making. In this regard, the objective of this chapter is to provide a detailed survey of various pregnancy-related abnormalities; and to explore various machine learning algorithms to classify/predict pregnancy-related abnormalities with higher accuracy. A generic framework that focuses more on classifying various features into normal and abnormal, and to be monitored patients to provide support and care during an emergency.