Accurate assessment of journal impact is essential for informing and guiding journal development. Existing journal normalized impact indicators are predominantly constructed at the field level. With increasing interdisciplinary integration and blurred disciplinary boundaries, the growing diversity of topics has rendered field-level normalized indicators insufficient for fine-grained journal impact evaluation. To address this, we previously proposed the Journal Normalized Impact (JNI), a topic-level normalized indicator that integrates topic modeling and citation data. However, JNI has limitations in topic clustering, the rationality of its calculation, and topic-level impact interpretation. This study proposes an improved framework and develops the series indicators, employing in-depth semantic topic modeling approach with z-score normalization and applying a “filter-classify-unify normalization” approach to ensure the robustness and interpretability of impact measurement. Empirical analysis confirms that the indicators effectively capture both overall and topic-specific journal impact. Compared to previous indicators, the indicators improve evaluative precision, offer more robust and stable measurements, and reveal more nuanced topic-level insights. We hope this study provides a foundation for refined, topic-based journal evaluation and contributes to more accurate and reliable research assessment.
扫码关注我们
求助内容:
应助结果提醒方式:
