Answers are grounded in this paper and cite their sources.
Analysis reveals limitations of LLMs in crafting book advertisements, indicating adjustments may enhance outcomes.
In China, the influence of large language models (LLMs) in text creation has been growing rapidly. However, some of them may not reach users’ expectations. Taking China's publishing industry as an entry point, this study systematically explores the practical application of LLMs. Four of China's well‐known LLMs were selected. The dataset was constructed by collecting the advertising texts of the top 100 books on the 2024 sales ranking of Dangdang.com (Top1 book‐selling platform in China), followed by data annotation by two publishing majors’ postgraduates. The study reveals that China's LLMs have shortcomings in capturing the core selling points of books and identifying users’ pain points, as well as inaccuracies in text emotion analysis and linguistic style. This study proposes optimization strategies that incorporate the Pain Point–Selling Point–Solution–Evidence logic framework into prompt engineering, along with fine‐tuning the model with emotional vocabulary data.
No takes yet. Share an insight, caveat, or question.
Zhai et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: