HENSOPT: a hybrid ensemble summarization framework with optimization-driven dynamic weighting and frequency-aware attention
Aposta: Revista de Ciencias Sociales · 28.05.2026 · 24 (112) · e894
DOI: 10.23900/ra.v24i112.894 ↗Kurz gefasst
Fachartikel in der Zeitschrift Aposta: Revista de Ciencias Sociales, online erschienen am 28.05.2026. Verfasst von Kancharla Bharath Reddy und 1 weiteren Autor:innen.
Abstract (Originalsprache)
AbstractThis study introduces a new hybrid ensemble summarization framework dubbed HENSOPT to achieve succinct and faithful summaries for long legal documents. The framework is a combination of both extractive summarization and abstractive summarization, namely the Legal Extractor (LEX) and the Legal Abstractive Refiner (LAR). The weights of the transformer are further optimized by Lion Optimization Algorithm (LOA) and the convergence is enhanced, and the Slime Mold Algorithm (SMA) is dynamically adjusted according to the quality of the summary. A frequency-aware attention mechanism is built in to give greater weight to rare but relevant words, minimizing semantic loss and ensuring factual consistency. This model is tested on LawSum, IN-Abs, and BillSum datasets, and metrics used to evaluate the model are ROUGE-1, ROUGE-2, ROUGE-L, BLEU and F1-score. Experimental results indicate that HENSOPT outperforms the baseline models, including BERTSum, PEGASUS and T5 Transformer, with significant improvements, such as gaining 7.8 percentage points on LawSum dataset when measured by ROUGE-1. The results validate the proposed framework's effectiveness for long-document summarization, its versatility in handling legal document types, computational efficiency while optimizing, and acceptability for large-scale correct, coherent, and topic-aware summarization applications.
Zitieren
APA
Reddy, K. B., & Jayabharathy (2026). HENSOPT: a hybrid ensemble summarization framework with optimization-driven dynamic weighting and frequency-aware attention. Aposta: Revista de Ciencias Sociales, 24(112), e894. https://doi.org/10.23900/ra.v24i112.894
Vancouver
Reddy KB, Jayabharathy. HENSOPT: a hybrid ensemble summarization framework with optimization-driven dynamic weighting and frequency-aware attention. Aposta: Revista de Ciencias Sociales. 2026;24(112):e894. doi:10.23900/ra.v24i112.894
BibTeX
@article{Reddy2026,
author = {Reddy, Kancharla Bharath and Jayabharathy},
title = {HENSOPT: a hybrid ensemble summarization framework with optimization-driven dynamic weighting and frequency-aware attention},
journal = {Aposta: Revista de Ciencias Sociales},
year = {2026},
volume = {24},
number = {112},
pages = {e894},
doi = {10.23900/ra.v24i112.894}
}
EUVIMED