TEXT AND VIDEO
SUMMERIZATION

This project focused on developing an abstractive summarization system using a seq2seq model in PyTorch, designed to condense large-scale YouTube video transcripts into precise summaries. By incorporating attention mechanisms, the system achieved a 15% improvement in BLEU scores and a 20% boost in ROUGE metrics. Advanced text preprocessing techniques like tokenization, contraction mapping, and stop-word filtering enhanced model accuracy by 25%, enabling grammatically coherent and concise summaries. This project demonstrates expertise in NLP, deep learning, and large-scale data processing.