publications

2026

1. SemEval

   AI-Monitors at SemEval-2026 Task 4: A Hybrid Embedding and LLM Ensemble for Narrative Similarity

   Vishnu Tripathi, Prakhar Joshi, Pragyananda Sahoo, and 4 more authors

   In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026), Jul 2026

   Abs Bib

   Narrative similarity requires reasoning over the deeper structural properties of stories - shared themes, causal progression, and outcomes - rather than surface-level lexical overlap. We describe AI-Monitors, our system for SemEval-2026 Task 4 (Track A), which determines which of two candidate stories is more narratively similar to a given anchor. We explore a progression of approaches - from embedding-based similarity to structured LLM prompting and ensemble construction - guided by four hypotheses about where narrative reasoning gains can be found. The final system achieves 75% test accuracy on 400 instances, ranking 3rd out of 47 systems and approaching the individual human annotator ceiling of 78%. Our key findings are: i) structured few-shot prompting substantially outperforms dense embedding similarity; ii) selecting ensemble components by how differently they make errors - rather than by accuracy alone - produces stronger predictions; and iii) how you describe an example to the model affects its predictions.

   @inproceedings{tripathi2026ai,
     title = {{AI}-Monitors at {S}em{E}val-2026 Task 4: A Hybrid Embedding and {LLM} Ensemble for Narrative Similarity},
     author = {Tripathi, Vishnu and Joshi, Prakhar and Sahoo, Pragyananda and Kumar, Gaurav and Arora, Piyush and Mani, Neel and others},
     booktitle = {Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026)},
     publisher = {Association for Computational Linguistics},
     month = jul,
     year = {2026},
     pages = {2139--2148}
   }

2. MBCC

   Marma Point Detection and Personalized Ayurvedic Diagnosis Using Computer Vision

   Pragyanand Sahoo, Neel Mani, Shastri Nimmagadda, and 1 more author

   In Indian Knowledge System and Wellbeing: Proceedings of MBCC 2025, Volume 1, Jun 2026

   Abs Bib

   Mental health is a global concern, with many individuals suffering from dysfunction due to depression, anxiety, common mental disorders, daily stressors, alcohol and substance use issues, and psychoses. Marma therapy addresses both psychological and physical challenges using contemporary methods. This traditional Indian Ayurvedic practice involves treating specific areas of the body. Marma therapy effectively reduces stress, anxiety, sadness, fears, and phobias, as supported by numerous studies. We gather images of users and analyse them based on various characteristics, including body type (Vata, Pitta, Kapha), height, current symptoms, structure, and medical history. This study outlines a comprehensive methodology for providing personalised recommendations for marma points to individuals. This solution could significantly impact Ayurveda and mental health.

   @incollection{sahoo2025marma,
     title = {Marma Point Detection and Personalized Ayurvedic Diagnosis Using Computer Vision},
     author = {Sahoo, Pragyanand and Mani, Neel and Nimmagadda, Shastri and Das, Amit},
     booktitle = {Indian Knowledge System and Wellbeing: Proceedings of MBCC 2025, Volume 1},
     publisher = {Springer Nature},
     month = jun,
     year = {2026},
     pages = {228}
   }

2025

1. COLING

   AI-Monitors at GenAI Detection Task 1: Fast and Scalable Machine Generated Text Detection

   Azad Singh, Vishnu Tripathi, Ravindra Kumar Pandey, and 6 more authors

   In Proceedings of the 1st Workshop on GenAI Content Detection (GenAIDetect), Jan 2025

   Abs Bib HTML PDF

   We describe the work carried out by our team, AI-Monitors, on the Binary Multilingual Machine-Generated Text Detection (Human vs. Machine) task at COLING 2025. This task aims to determine whether a given text is generated by a machine or authored by a human. We propose a lightweight, simple, and scalable approach using encoder models such as RoBERTa and XLM-R. We provide an in-depth analysis based on our experiments. Our study found that carefully exploring fine-tuned parameters such as i) no. of training epochs, ii) maximum input size, iii) handling class imbalance etc., plays an important role in building an effective system to achieve good results and can significantly impact the underlying tasks. We found the optimum setting of these parameters can lead to a difference of about 5-6% in absolute terms for measure such as accuracy and F1 measure. The paper presents crucial insights into optimal parameter selection for fine-tuning RoBERTa and XLM-R based models to detect whether a given text is generated by a machine or a human.

   @inproceedings{singh-etal-2025-ai,
     title = {{AI}-Monitors at {G}en{AI} Detection Task 1: Fast and Scalable Machine Generated Text Detection},
     author = {Singh, Azad and Tripathi, Vishnu and Pandey, Ravindra Kumar and Saho, Pragyanand and Joshi, Prakhar and Mani, Neel and Alagh, Richa and Mishra, Pallaw and Arora, Piyush},
     booktitle = {Proceedings of the 1st Workshop on GenAI Content Detection (GenAIDetect)},
     editor = {Alam, Firoj and Nakov, Preslav and Habash, Nizar and Gurevych, Iryna and Chowdhury, Shammur and Shelmanov, Artem and Wang, Yuxia and Artemova, Ekaterina and Kutlu, Mucahid and Mikros, George},
     month = jan,
     year = {2025},
     address = {Abu Dhabi, UAE},
     publisher = {International Conference on Computational Linguistics},
     pages = {230--235},
   }