Bibliometrics effects of a new paper level classification

Area of Science:

• Bibliometrics
• Scientific Document Classification
• Information Science

Background:

• Traditional journal-based classification systems, like Scopus's All Science Journal Classification (ASJC), assign documents to predefined categories.
• These systems may not accurately reflect the interdisciplinary nature of modern research or the specific context of citations.

Purpose of the Study:

• To comparatively analyze a Scopus journal-based classification system (fractional model) against a novel item-by-item classification system.
• To evaluate the impact of citation origin on document classification and bibliometric indicator performance.

Main Methods:

• The study adapted the Scopus journal-based assignment method to a fractional model.
• A second system was developed using an item-by-item classification based on reclassified references by citer origin.
• Comparisons were conducted at Scopus area and category levels, alongside bibliometric indicator analysis.

Main Results:

• The paper-level system reduced the number of assigned categories per document, yielding higher single-category assignments than ASJC.
• Reclassification using the paper-level system accentuated differences in category sizes, enlarging the largest and shrinking the smallest.
• The item-by-item system produced more homogeneous distributions in normalized impacts and adjusted excellence values more uniformly.

Conclusions:

• The paper-by-paper classification system offers a more nuanced and potentially accurate method for categorizing scientific documents.
• This approach provides a more refined analysis of bibliometric indicators, particularly normalized impacts and excellence metrics.
• The findings suggest advantages of item-level classification for understanding research impact and interdisciplinary connections.