An Efficient Hybrid Model to Summarize the Text Using Transfer Learning

In today’s era, an enormous quantity of information is generated everyday by the Internet and this information need a specific tool by which we can get a more useful and important part of that information. In this study, we propose a design to generate a short and coherent version of a longer document. We can achieve this by two methods first one is extractive means it will take out the important words from the same document and the second one is abstractive means firstly it will understand the semantics of the paragraph and then it will generate a new text summary. This study work on the hybrid model which can attain this task by distilling salient words from the document and by generating completely new sentences by understanding the semantics. We used T5 which is a transfer learning model and get a better ROUGE-1 score of 45.2, ROUGE-2 score of 23.1, and ROUGE-L score of 39.7, which is better than all other models.