SNSudarshan Nath

Sudarshan Nath

*This is an algorithmically generated author record based on publicly available information.

Sudarshan Nath has a total of 15 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    A Hybrid Transformer-Driven and Embedding-Augmented Framework for Automated Contextual MCQ Generation with Semantic Distractor Engineering and Difficulty-Aware Classification

    by Sudarshan Nath, Suparna DasGupta, Soumyabrata Saha

    This study introduces an unsupervised technique for creating multiple-choice questions (MCQs) and deletions from an AI-related article corpus. The method is trained with an MCQ generator via a new strategy that utilizes natural language processing (NLP) principles to generate high-quality, contextually appropriate questions automatically. The main aim is to enhance learning tests and support the learning process by generating pedagogically sound MCQs from various textual sources. The process begins by abstracting the input texts using a recent methodology that extracts useful keywords, bigrams, and unigrams and stores them within a structured database. The keywords are then weighted with inverse document frequency (IDF), while distractor options are generated through a context-sensitive similarity method through paradigmatic association discovery. The generation of questions such that the information would be complete and meaningful were ensured by the exclusion of sentences with discourse connectives during the question formulation phase. It was tested on four different text passages and achieved an average accuracy rate of 85%, according to human experts' estimates. In comparison to other existing methods, our approach has a better level of accuracy and no human intervention during generation, thus making it extremely efficient and scalable. The outcomes show an enhanced quality and relevance in the generated questions. The technique has wide-ranging applications, including being applied in massive open online courses (MOOCs), intelligent tutoring systems, and self-testing tools for students. Even though the system was initially conceived for AI-related content, it can find applications in other domains of application as well.

  • TriFusionRank: An AI Based Multi-Factor Sentence Scoring Framework for Lightweight Extractive Summarization

    by Sudarshan Nath, Suparna Dasgupta, Soumyabrata Saha

    Extractive summarization is the key to alleviating the problem of information overload by choosing essential sentences from a document and creating short summaries. In this paper, a new extractive summarization method that combines three different scoring mechanisms, including term frequency-inverse document frequency (TF-IDF), sentence positional weighting, and semantic similarity with the document title, is proposed. In contrast to traditional techniques based on statistical or graph-based models, our method utilizes a hybrid approach to best capture sentence salience and contextual utility. Experimental evaluations on benchmark data show that the proposed method has competitive ROUGE scores against strong baselines while being interpretable and of low computational cost. This method provides a light but efficient solution for real-time summarization operations and provides a basis for future developments that include semantic comprehension and deep learning integration.

  • Open Access

    Context-Aware Predictive Optimization of Non-Functional Requirement Conflicts in DevOps Using Machine Learning

    by Anirban Bhar, Suparna Dasgupta, Soumyabrata Saha, Amrut Ranjan Jena, Koushik Karmakar, Shubhendu Banerjee, Pushpita Roy, Bikas Mondal, Soumya Bhattacharyya, Sudarshan Nath, et al.

  • Open Access

    An Investigation of Suicidal Ideation from Social Media Using Machine Learning Approach

    by Soumyabrata Saha, Suparna Dasgupta, Adnan Anam, Rahul Saha, Sudarshan Nath, Surajit Dutta

    Despite improvements in the detection and treatment of severe mental disorders, suicide remains a significant public health concern. Suicide prevention and control initiatives can benefit greatly from a thorough comprehension and foreseeability of suicide patterns. Understanding suicide patterns, especially through social media data analysis, can help in suicide prevention and control efforts. The objective of this study is to evaluate predictors of suicidal behavior in humans using machine learning. It is crucial to create a machine learning model for detection of suicide thoughts by monitoring a user's social media posts to identify warning signs of mental health issues. Through the analysis of social media posts, our research intends to develop a machine learning model for identifying suicide ideation and probable mental health problems. This study will help immensely to comprehend the environmental risk factors that influence suicidal thoughts and conduct across time. In this research the use of machine learning on social media data is an exciting new direction for understanding the environmental risk factors that impact an individual's susceptibility to suicide ideation and conduct over time. The machine learning algorithms showed high accuracy, precision, recall, and F1-score in detecting suicide patterns on social media data whereas SVM has the highest performance with an accuracy of 0.886.