MASF: A Multi-Model Adaptive Selection Framework for Abstractive Text summarization
Researchers have developed a Multi-Model Adaptive Summarization Framework (MASF) to enhance abstractive text summarization. This framework integrates multiple fine-tuned transformer models, each generating a summary for a given article. An adaptive selection mechanism then chooses the best summary based on lexical similarity and semantic relevance metrics. MASF demonstrated superior performance, achieving the highest BERTScore of 88.63% and outperforming models like GPT3-D2, Falcon-7b, and Mpt-7b on the CNN/DailyMail dataset. AI
IMPACT This framework could improve the quality and consistency of automated text summarization across diverse content types.