CLOF: A Coupled Framework for Quantifying Research Impacts via A Double Iterative Power Method
Year: 2026
Type: article
Abstract: Research impact quantification plays a crucial role in academic assessment. Conventional citation-based metrics have several limitations in fair and comprehensive research evaluation. They often fail to recognize foundational work that receives limited early attention. They may also overlook sustained influence over time, differences across research fields, citation quality reflected by citation contexts and the expertise of citing authors, and variations in author contributions. We incorporate these aspects at the paper level through four factors: originality-oriented downstream influence, long-term influence, field-specific ranking, and citation quality. At the researcher level, we quantify credit according to author order. Based on these, we propose CLOF, a unified metric framework that quantifies the impacts of papers and researchers in an interdependent and iterative manner. To reduce reliance on manually specified weighting schemes, we employ multiple large language model (LLM)-based paper scorers. Their outputs are aggregated using an α-trimmed procedure to generate robust global signals for learning the combination weights. These learned weights further guide the coupled iterative updates of paper and researcher scores via a Double Iterative Power (DIP) method with theoretical convergence guarantees. Extensive experiments on academic datasets show that CLOF uncovers influence patterns across papers and researchers that are not captured by existing impact metrics.
Institutions: City University of Hong Kong, University of Michigan
Language: English
Subfield: Library and Information Sciences
Field: Social Sciences
Domain: Social Sciences
Open Access status: diamond
OpenAlex ID: W7236287569
