The amount of data generated through online has increased rapidly due to the emergence of the Internet and the rapid progress of social media. This accumulated data offers unprecedented opportunity to different organizations for analyzing the view or opinion available as reviews toward any product, article, or any other entity, which aids in efficient development of business strategies as well as decision making. However, this humongous data available in an unstructured manner and in natural language, poses a great challenge for extracting any significant information in text. Thus, to obtain the sentiment of the users as well as concise representation of the text, sentiment rating prediction and abstractive summarization is essential. Hence, a novel method for extracting the abstractive summary from reviews using CNN-transfer learning is devised for predicting sentiment rating using a hybrid deep learning network comprising deep neuro fuzzy network (DNFN) and deep maxout network (DMN), for estimating the sentiment rating. Further, the techniques are evaluated for their effectiveness in terms of precision, recall, f-measure, and ROUGE rouge measure. The developed model showed superior results compared to existing methods for abstractive summarization.