Factors associating with or predicting more cited or higher quality journal articles: An Annual Review of Information Science and Technology (ARIST) paper
这篇综述梳理了文章文本和元数据中与更高被引或更高质量相关的因素，并回顾了用统计回归或机器学习预测引文数量或文章质量的研究，发现关联强度和方向因领域而异，缺乏系统性规律和因果证据。
Abstract Identifying factors that associate with more cited or higher quality research may be useful to improve science or to support research evaluation. This article reviews evidence for the existence of such factors in article text and metadata. It also reviews studies attempting to estimate article quality or predict long‐term citation counts using statistical regression or machine learning for journal articles or conference papers. Although the primary focus is on document‐level evidence, the related task of estimating the average quality scores of entire departments from bibliometric information is also considered. The review lists a huge range of factors that associate with higher quality or more cited research in some contexts (fields, years, journals) but the strength and direction of association often depends on the set of papers examined, with little systematic pattern and rarely any cause‐and‐effect evidence. The strongest patterns found include the near universal usefulness of journal citation rates, author numbers, reference properties, and international collaboration in predicting (or associating with) higher citation counts, and the greater usefulness of citation‐related information for predicting article quality in the medical, health and physical sciences than in engineering, social sciences, arts, and humanities.