Marathi Text Document Summarization using Neural Networks

Abstract

The internet is comprised of web pages, news articles, status updates, blogs, and much more. It is difficult to navigate through this data as it is unstructured and usually discursive. Manual summarization of large documents of texts is tedious and error-prone. Also, the results in such kind of summarization may lead to different results for a particular document. Thus, Automatic text summarization has become important due to the tremendous growth of information and data. It chooses the most informative part of the text and forms summaries that reveal the main purpose of the given document. It yields a summary produced by a summarization system which allows readers to comprehend the content of the document instead of reading every individual document. So, the overall intention of Text Summarizer is to provide the meaning of the text in fewer words and sentences. To perform extractive text summarization, we propose to use a Recurrent Neural Network (RNN) – a type of neural network that can perform calculations on sequential data (e.g. sequences of words) – as it has become the standard approach for many Natural Language Processing tasks.