Tamil Text Summarization Using Graph Convolutional Networks And Google’s MT5 Transformer

Text summarization in Tamil language is an important problem in natural language processing and understanding and in information retrieval. Automatic text summarization is getting more attention nowadays as it helps in saving time in decision making by conveying the information concisely. The primary objective of this experimental project is to propose a methodology to address the problem of summarization for Tamil language text documents, which can generate concise meaningful summaries using Graph Convolutional Network (GCN) and Google Multilingual Text-to-Text-Transfer-Transformer (Google mT5). GCN is used to extract important sentences for extractive summarization and Google mT5 to summarize the extracted sentences, which will give a concise summary of the input text document without losing the meaning of the documents. Experiments are carried out using Huggingface datasets, and the ROUGE toolkit is used to evaluate the precision, recall, and F1 score of summaries generated by the system and also human evaluation.