Abstract: At present, the output of academic papers grows with each passing day. It is a main task to be solved urgently that researchers should quickly find the achievements with the high degree of innovation among the multitudinous academic papers. After elaborating the theoretical basis of the model, a comprehensive BJRC model of academic achievements’innovation degree is constructed from four aspects: accumulation basic(B), carrier’s impact factor(J), recognition(R) and context similarity(C). And then the number of times that "artificial intelligence" papers cited by patents from 2015 to 2017 are used for modelling vertification. Additionally, this paper makes a comparative analysis from two perspectives: internal comparisons and external comparisons with other mainstream models for innovation. The result proves that BJRC model is not only scientific and reasonable, but might also be used practically, which conducive to identify the most and some highly innovative academic papers. The BJRC model constructed in this paper could be extended to measure the degree of innovation of a certain branch of learning or larger research area in the future.

Key words: Innovation degree, Author accumulation, Carrier’s impact Factor, Recognition, Context similarity, Paper evaluation