A Hybrid Approach for Multimedia Summarization

Text summarization is the basic method in NLP (Natural Language Processing) application which makes the text data into a shorter version. All over the world a lot of multimedia data is generating which includes text, audio, video, image, etc., Generating the short summary from entire text will reduce our task of understanding the things. In the proposed work, film reviews data is taken to perform summarization. A film review is an article that reviews and analyses a film and is published in a magazine, newspaper, video, or scholarly journal. As a result, reading through the entire conversation will take a long time to comprehend the reviews and consider the film. To perform text summarization, we need to work on multi-media data to generate the short summary of the entire content. By scanning the images, audio, videos and converting them into text documents, multi-media data will reduce the difficulty for the machine to understand the important information from it. In this project, proposing a Hybrid (combination of Extractive summarization and Abstractive summarization) multi-media summarization method that uses NLP techniques, image processing, speech processing to retrieve the information contained in multimedia model. The proposed work includes the sequence of implementations namely, Multimedia processing, applying NLP techniques, Seq-to-Seq modelling to the text, and performing sentiment analysis using a Naive Bayes classifier.