Sentiment analysis (SA) is commonly applied to digital textual data, revealing insight into opinions and feelings. Many systematic reviews have summarized existing work, but often overlook discussions of validity and scientific practices. Here, we present an overview of reviews, synthesizing 38 systematic reviews, containing 2,275 primary studies. We devise a bespoke quality assessment framework designed to assess the rigor and quality of systematic review methodologies and reporting standards. Our findings show diverse applications and methods, limited reporting rigor, and challenges over time. We discuss how future research and practitioners can address these issues and highlight their importance across numerous applications.

翻译：情感分析（SA）常被应用于数字文本数据，揭示观点与情感的内在信息。众多系统综述已对现有工作进行总结，但往往忽视了对有效性与科学实践规范的讨论。本文提出一种综述之综述，综合了38篇系统综述，涵盖2,275项原始研究。我们设计了一套定制化的质量评估框架，用于评估系统综述方法论与报告标准的严谨性及质量。研究结果表明：应用领域与方法呈现多元化，但报告严谨性有限，且存在长期性挑战。我们探讨了未来研究与实践者如何应对这些问题，并强调其在众多应用场景中的重要性。