This document presents a comparative analysis of methods for abstractive text summarization, highlighting their pros and cons, which focused on the use of deep learning method in abstractive text summarization. Nowadays, Internet is used widely throughout the world, due to which an enormous amount of data is created on the Internet through news, product reviews, research papers, emails, and so on. It is very difficult for users to read the whole document in a limited time period. Automatic text summarization comes into focus to solve this problem by creating summary automatically. Text summarization is classified into two types, i.e., extractive and abstractive text summarization. Abstractive text summarization is further classified into structured-based and semantic-based approaches. The concept of these methods has been explored along with some known features of extractive text summarization which introduced deep learning techniques. LSTM model of recurrent neural networks with sequence to sequence have given effective results. In this paper, the proposed system includes a BiLSTM model, i.e., bidirectional LSTM with attention mechanism to achieve better results.