The NLP Techniques for Automatic Multi-article News Summarization Based on Abstract Meaning Representation

Abstract

The analysis of natural language texts is one of the most important knowledge discovery tasks for any organization. Automated text summarization systems can reduce the size of the text while keeping the important part of the text and desired information. The applications of summarization are summaries of email thread, action items from a meeting, simplifying text by compressing sentences and abstracts of any document, an article, etc. It comes in two ways, single-document summarization and multiple-document summarizations. In single-document summarization technique, given a single document produces abstract, outline. Whereas, with multiple-document summarization technique, given a group of documents produce a list of the content such as a series of news stories on the same event or a set of web pages about some topic or questions. Consequently, there are two ways of doing summarization, an extractive summarization creates the summary from phrases or sentences in the source document and an abstractive summarization express the ideas in the source documents using different words. However, the abstractive summarization methods are very comprehensive to get the abstract meaning of multiple articles and generate the summary. Thus, the text contents are analyzed and extract named entities using Stanford NER tool in different aspects to get abstract meaning of multiple articles. In this study, an abstract summary of article using named entities are presented.