Development of Summarization Tool for Novels
B Dhanushree, V Megha, Aditi Sharma, S Ananya, C Deepthi · 2025
Summarization tools are pivotal for closing the gap to an overbearing body of textual data; however, they generally address questions arising from simpler forms such as news articles, leaving the novel's faculties mostly untouched. Summarizing novels increases the challenges faced on the summation frontelaborate plotting twists, layered character arcs, and richness of the text. Added to that, novel descriptions require an explanation that provides a great plight and sustains the artistic content-one dimension where laid-back methodologies may often fall short. Put simply, long-format literature interlinks themes and emotions amid a coiling list of subplots, thus emphasizing how difficult it is to summarize well without falling into detail loss. This newly conceived novel summarization tool couples a BERT model with TextRank-like algorithms. It outputs contextual embeddings from BERT to pull sentences demonstrating central themes and the evolution of the plot while putting the added algorithms to work to fulfill various user requests, including hierarchical summaries. Many such outputs make the system friendly towards casual readers and researchers alike, providing a summary with an upper bend towards conciseness alongside greater detailed information. Although promising, the challenges still lie in reaching some equilibrium between narrative coherence and flexibility, especially whilst challenged by non-linear storytelling, stylistic nuance, and varying expectations of readers. Nevertheless, qualitative and quantitative analyses show the meaningful ways in which the system performs well, enabling readers and researchers to trek further toward literature while providing additional input to the much wider fields of NLP and literary analysis.