Neural Abstractive Text Summarization with Sequence-to-Sequence Models

In the past few years, neural abstractive text summarization with sequence-to-sequence (seq2seq) models have gained a lot of popularity. Many interesting techniques been proposed to improve seq2seq models, making them capable handling different challenges, such as saliency, fluency and human readability, generate high-quality summaries. Generally speaking, most these differ in one three categories: network structure, parameter inference, decoding/generation. There are also other concerns, efficiency parallelism for training model. this article, we provide comprehensive literature survey on from viewpoint structures, strategies, summary generation algorithms. Several were first language modeling tasks, machine translation, later applied summarization. Hence, brief review models. As part survey, develop an open source library, namely, Neural Abstractive Text Summarizer (NATS) toolkit, An extensive set experiments conducted widely used CNN/Daily Mail dataset examine effectiveness several components. Finally, benchmark two implemented NATS recently released datasets, Newsroom Bytecup.