Abstractive Summarization Using Categorical Graph Network

The rapid development of technologies produce enormous amount data which have lot hidden insights. Extracting these insights are challengeable for researchers and industrialists. Most the in textual unstructured format. Text mining is prominent research area that has being utilized analysis. Document summarization an effective application provides summary given content. This work mainly focused on generating abstractive from multiple documents. It contributes using cate-gorical graph network. Lot duplicate or redundant sentences there Proposed CATSum, a based technique identifies duplication similarities sentences. proposed used ALBERT encoder model to train datasets. Then it built content connection between measured ROUGE-1, ROUGE-2 ROUGE-L metrics produced better accuracy than baseline methods.