Text Summarization for Personalized Movie Review

Abstract

The aim of this project is to create a personalised review creation system for movies using a sequence-to-sequence model and abstractive text summary techniques. The idea is to developinsightful and personalised evaluations that capture the spirit and sentiment of a film.Abstractive text summarization is an Natural language processing (NLP) techniqueof generating new and brief summary of source text data. It generates new lines in summary of text data which are relevant to the original lines. Abstractive summarization yields a numberof applications in different domains, from books and literature, to scienceand R&D, to financial research and legal documents analysis. This project introduces a fresh method for generating personalized movie reviews using abstractive text summarization. The approach employs a BERT-based encoder-decoder model, trained onuser reviews enriched with movieratings. This enables the model to create reviews that are not only informative and captivating but also match the user's preferences. By merging NLP and user preferences, we provide summaries that offer insights into reviewswhile aligning with the user's cinematic likes. In the era of abundant online content, our personalized summarization technique is a valuable tool for helping users navigate movie-related information. Keywords : Transformers, Bert model, Bert for Sequence Classification, Bert Tokenizer, Pegasus model,Pegasus tokenizer, Data Loader, Pandas.