• Adarsh Gupta
• Afreen Sultana
• Ayush Singh Tomar
• Digvijay Khanapurkar
• Ms. Visalini

Text summarization can be used by personal or specialised assistants. Apart from that it can be used for many personalized devices or applications like mail clients, report generation, news feed etc. There is a problem of searching for relevant documents from the number of documents available, and observing relevant information from it. Earlier there was a technique called Extraction in which summary of document was generated by the words which is already present in the document, no new words were added. But ABS uses a new method called Abstraction which adds new words to form a summary. A long text needs to be summarized, so that human can understand it easily. A method called novel sentence-level policy gradient used by ABS to bridge the non differentiable computation between two neural networks in a hierarchical way. ABS first selects salient sentences and then rewrites them abstractively to generate a concise overall summary. To generate a concise overall summary, they are not restricted to simply selecting and rearranging passages from the original text. Using CNN/Daily Mail dataset ABS achieves the way to find for significantly higher abstractiveness scores. It first operates at the sentence-level and then the word-level. ABS enables parallel decoding of neural generative model that results in substantially faster inference speed.

Abstraction. Extraction, Text Summarization

"BE SPECIFIC : ABSTRACTION BASED SUMMARIZATION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 5, page no.166-169, May-2019, Available :http://www.jetir.org/papers/JETIRCD06028.pdf

"BE SPECIFIC : ABSTRACTION BASED SUMMARIZATION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 5, page no. pp166-169, May-2019, Available at : http://www.jetir.org/papers/JETIRCD06028.pdf