Barkavi Sundararajan
About
I am a final-year PhD student in Computing Science at the University of Aberdeen under the supervision of Dr. Yaji Sripada and Prof. Ehud Reiter. My research focuses on improving factual accuracy (how faithfully the summaries are grounded in the input data) and reducing hallucinations in data-to-text generation by analysing input quality and structure. I also refined a manual error annotation protocol for factual errors and am translating it into a structured, semi-automated evaluation framework to complement human evaluation.
Publications
Barkavi Sundararajan, Somayajulu Sripada, and Ehud Reiter (2025). Input Matters: Evaluating Input Structure’s Impact on LLM Summaries of Sports Play-by-Play. Proceedings of the 18th International Natural Language Generation Conference (INLG), Hanoi, Vietnam, October 2025 - [PDF]
Barkavi Sundararajan, Somayajulu Sripada, and Ehud Reiter (2024). Improving Factual Accuracy of Neural Table-to-Text Output by Addressing Input Problems in ToTTo. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL main), Mexico - [PDF]
Craig Thomson, Ehud Reiter, and Barkavi Sundararajan (2023). Evaluating factual accuracy in complex data-to-text. Computer Speech & Language (CSL) - [PDF]
Barkavi Sundararajan, Somayajulu Sripada, and Ehud Reiter. (2022) Error Analysis of ToTTo Table-to-Text Neural NLG Models. Proceedings of the Second Workshop on Natural Language Generation, Evaluation, and Metrics (GEM), Abu Dhabi) - [PDF]
News and Updates
- Dec 2025: Awarded Scottish Informatics and Computer Science Alliance (SICSA) PhD Travel Funding.
- Nov 2025: Presented my paper at the INLG 2025 Conference, Hanoi, Vietnam.
- Oct 2025: Presented an overview of my doctoral research at the Young Researchers in Natural Language Generation (YNLG) Workshop, Hanoi, Vietnam.
- Jun 2025: Presented my Turing internship work, Unlocking and Codifying Free Text into Structured Insights for Justice Outcomes, at the Data First Showcase, London.
- Nov 2024: Started a part-time Data Science Turing Internship at the UK Ministry of Justice.
- Jul 2024: Presented my research work at the SICSA 2024 Conference, Aberdeen, and served as a Student Volunteer.
- Jun 2024: Presented my research work at the NAACL 2024 main conference, Mexico, and served as a Student Volunteer.
- Oct 2022 – Jul 2024: Teaching Assistant for MSc AI courses: Natural Language Generation, Data Mining with Deep Learning, and Evaluation of AI Systems.
- Sep 2023 – Oct 2023: Built a Retrieval-Augmented Generation (RAG) proof-of-concept with a development team and presented it to an industry client (Bray Oil & Gas).
- Dec 2022: Presented my research work at the second Workshop on Generation, Evaluation, and Metrics (GEM), Abu Dhabi.
- Nov 2020 – Apr 2021: Worked as a Full Stack Software Developer Intern at James Fisher AIS, Aberdeen.
- Jan 2021: Graduated with a Master of Science in Information Technology from the University of Aberdeen, with distinction.
- Aug 2011 – Dec 2019: Worked on UK and European B2B telecom projects at Plintron Mobility Solutions, Chennai, India. Delivered technical solution proposals, supported GDPR and other EU regulatory compliance, and developed strong telecom product expertise; left the role to pursue postgraduate studies in Computing Science.
- Mar 2009: Presented Pattern analysis using Neural Networks with Back Propagation algorithm for recognising fingerprints and handwritten digits in National level Technical Symposium, Chennai.