A multi-dimensional indicator system for identifying highly innovative papers: a knowledge absorption, creation, and diffusion perspective

Conventional metrics often emphasize a single aspect of academic innovation, such as citation counts, topical novelty, or disruption, and therefore fail to capture the multi-mechanism processes through which academic innovation emerges and spreads. This study develops a multi-dimensional indicator system grounded in a knowledge-evolution perspective that conceptualizes innovation across three stages: knowledge absorption, creation, and diffusion. Each stage corresponds to distinct, measurable traces in citation networks and textual semantics. Furthermore, the framework integrates citation- and content-based indicators and refines topic-novelty and knowledge-disruption measures through semantic and temporal modeling to reduce granularity and aggregation bias. Empirical validation is conducted using 1940 publications by Price Medal recipients and their cited and citing records (54,045 unique documents). Robustness is further examined through inter-indicator correlation analysis, comparison with citation impact, case validation using top-ranked innovative papers and representative works by Loet Leydesdorff, as well as a cross-field validation experiment. Results demonstrate that the proposed system effectively identifies highly innovative research and captures domain-specific patterns of knowledge generation and dissemination, offering a comprehensive and scalable approach for evaluating academic innovation.