Research Article
A Comprehensive Review of Generative AI Techniques for Tamil Natural Language Processing
@INPROCEEDINGS{10.4108/eai.28-4-2025.2357753, author={Kiruthika S S and Nalinipriya G}, title={A Comprehensive Review of Generative AI Techniques for Tamil Natural Language Processing}, proceedings={Proceedings of the 4th International Conference on Information Technology, Civil Innovation, Science, and Management, ICITSM 2025, 28-29 April 2025, Tiruchengode, Tamil Nadu, India, Part I}, publisher={EAI}, proceedings_a={ICITSM PART I}, year={2025}, month={10}, keywords={generative ai nlp llms llama 2 lora qlora multilingual translation domain-specific nlp bias mitigation cultural sensitivity}, doi={10.4108/eai.28-4-2025.2357753} }- Kiruthika S S
Nalinipriya G
Year: 2025
A Comprehensive Review of Generative AI Techniques for Tamil Natural Language Processing
ICITSM PART I
EAI
DOI: 10.4108/eai.28-4-2025.2357753
Abstract
Generative text AI revolutionized NLP by producing massive new advances in low-resource languages. To attain better efficiency at a low computational cost, this works has dedicated to raise the fine-tuning process of largescale language models, such as LLaMA 2, with the integration of techniques like Low-Rank Adaptation (LoRA) and Quantized LoRA (QLoRA). The large amount of code-switching that occurs in CoNLL-17’s Dravidian languages such as Tamil, a low resource, multilingual language, has made it an interesting case to tackle, given the emphasis on issues such as: dialectal variation, code-switching and data scarcity. A model-based vocabulary, a domain-specific dataset, and appropriate metrics have resulted in better performance from generative AI. Content creation for the Tamil language has benefited from utilization of the generative tools, namely cross-lingual adaptation and model combination.
