Citation structural diversity: a novel metric combining structure and semantics for literature evaluation

With the continuous expansion of research activities and the deepening of interdisciplinary studies, traditional bibliometric indicators are increasingly inadequate in reflecting the dissemination paths and multidimensional value of literature within the knowledge network. To address this challenge, this paper proposes a citation structural diversity metric that integrates both structural and semantic information. Built upon a citation network, the metric comprehensively considers the knowledge inheritance relationships between documents (bibliographic coupling) and their implicit structural associations (co-citation) to construct composite structural features, and incorporates semantic associations among documents to extract semantic features, thereby enabling a more precise depiction of a publication’s influence within the academic network and its interdisciplinary dissemination capability. Experimental results demonstrate that literature with higher structural diversity exhibits greater short-term citation activity and maintains strong academic influence in long-term citation trends. Through grouped statistical analysis and a ten-year longitudinal study, the method proves effective in uncovering the multidimensional value of scholarly works. Furthermore, in the context of interdisciplinary research, structural diversity is shown to be positively correlated with topic breadth, validating its utility in identifying interdisciplinary contributions. This metric offers a novel perspective for improving literature evaluation methods and highlights the unique role of structural diversity in assessing interdisciplinary knowledge diffusion. The code is available at https://github.com/mingyue15694/Citation-Structural-Diversity/tree/master .