Long Short-Term Memory Networks for Text Summarization using NLP and ML

In the digital world today, people are inundated with enormous amounts of text from a wide variety of sources including websites, newsletters, journals, blogs and legal documents. Reviewing and comprehending all this content can be tedious and mentally exhausting. In order to address this issue, our project will deliver an AI-based text summarization system that utilizes Natural Language Processing (NLP) and Machine Learning to summarize lengthy documents into brief summaries automatically via NLP techniques, which will produce highlights from text or complete sentences. The two forms of summarization include extractive summarization, where the system will extract key sentences from the source content, and abstractive summarization, where the system will summarize with its own (new) sentences while maintaining the understanding of the source content. Using these two forms of summarization techniques will lead to better summaries that are both factual and coherent. Our algorithms employ base NLP techniques and advanced models (e.g. BART, TF-IDF etc.) for the feature extraction combined with contextual understanding of the source text which will allow our summarization system to provide worthwhile summaries. Made for real time use, the tool is designed with a user friendly Gradio interface for easy access by a broad audience and does not require a technical background. Multiple input formats are enabled, including plain text, PDFs, audio, and video (converted to text) all of which makes it possible to utilize in many different fields of practice such as educational, media, business, and legal services. By automating the summarization process, this project will ideally save time and effort in going through information that is represented in larger amounts of text - ultimately allowing for productivity and decision-making to be possible.