ENHANCING THE QUALITY OF ACADEMIC PAPER ABSTRACTS USING LARGE LANGUAGE MODELS: A CASE STUDY ON "DIGITAL ECONOMY" PAPERS IN CHINA NATIONAL KNOWLEDGE INFRASTRUCTURE (CNKI)
Lin Zhong, ChaoMin Gao · Educational research and human development. · 2025
To evaluate the writing quality of academic paper abstracts and explore the applicability of large language models (LLMs) in abstract optimization, this study selects 5,054 papers on the topic of "Digital Economy" from CNKI as samples. A quantitative scoring analysis assesses the abstracts' performance in four dimensions: research objective, research methodology, research results, and research conclusions. Additionally, abstracts with significant deficiencies are regenerated using LLMs and subsequently evaluated. The data reveal that 57.44% of the abstracts fail to effectively summarize the core content of the research, with particularly pronounced issues in the descriptions of research methodology and results. Abstracts generated by LLMs exhibit excellent structural integrity, logical coherence, and linguistic conciseness. The findings indicate that academic paper abstracts in China have significant deficiencies in expressing research methodology, results, and conclusions, necessitating improvements through technological means. Given their strong capability in abstract writing, LLMs should be utilized to enhance the quality of academic abstracts.