Abstract: No consensus has been reached on the characteristics of breakthrough papers and the theoretical models for their identification. The content analysis method is adopted to analyze the similarities and differences between the characteristics of breakthrough papers and breakthrough research. A conceptual model of breakthrough paper identification is constructed from citation perspective and reflects the impact of breakthrough paper on the citation network and mainstream theories. On this basis, a breakthrough paper generation model is proposed. The influence of knowledge recombination on the originality of breakthrough papers is discussed from backward citation perspective. On the other hand, a breakthrough paper diffusion model is proposed to sort out the different paths of scientific impact cross-domain diffusion of breakthrough papers from forward citation perspective. It is found that breakthrough papers may not necessarily have all the characteristics of breakthrough research. The core characteristics consist of a huge influence on scientific research, challenging mainstream theories and interdisciplinary. The study defines the core characteristics of breakthrough papers. The conceptual model, generation model and diffusion model of breakthrough papers identification are constructed to provide a theoretical basis for the subsequent breakthrough paper identification research.

Key words: Breakthrough paper, Conceptual model, Generation model, Diffusion model, Citation perspective