Text Summarization Of Scientific Articles

The rapid growth of scientific publications, summarizing lengthy and complex articles has becomeincreasingly challenging. Traditional extractive summarization methods fall short in capturing the semanticcontent and relationships between scientific terms. This paper presents a abstractive summarization modelspecifically designed for summarizing scientific articles. The proposed model leverages SciBERT, a languagemodel pre-trained on scientific literature, and Page Rank algorithm to construct and encode a knowledge graphbased on extracted entities and relationships from the text. The model generates abstractive summaries bypreserving the hierarchical structure and semantic integrity of long scientific documents with help of NL