A deep learning based text summarization process

Abstract

Technological innovation has greatly improved our quality of life. But people's attention spans are getting shorter and people want to read for shorter periods of time at a rapid pace. Because of this, it's crucial to give a concise description of the most significant news item and the most logical summary that aligns with the synopsis in order to give a rapid review of the essential news or article. In this age of information, there is a vast amount of textual material at our disposal. Examples of sources include online documents, news stories, articles, and consumer reviews of various products and services. Summarizing texts is a technique for automatically summarizing any text, document, or paragraph. A summarized text is just the original material reduced to its most basic form. The primary goal of this effort is to provide a concise, easy-to-read summary that is both significant and comprehensible. since the primary barrier to communication is language. By offering a streamlined version of the material, text summary can assist in reducing the amount of time needed to read and comprehend lengthy publications. Text summaries can aid in highlighting key ideas and enhancing readers' understanding of the content as a whole. We have gathered information from the online portal Kaggle, which summarizes Amazon evaluations of products. We must apply our model in order to obtain an outline. Our model is a bi-directional RNN decoder and encoder with support for sequence-to-sequence, which uses an LSTM to provide the summary. We've encountered a number of issues with this project, including preprocessing, vocabulary counting, word embedding, and missing word counts. Our primary objectives for this project are to calculate the operational loss, create a fluid outline, and develop a more advanced technique for summarizing English texts. Our primary objective with this model was to create an abstractive summarizer based on our dataset