Abstract: 【Purpose/significance】There are many and complex methods for journal evaluation, and it is impossible to distinguish be‐tween good and bad, and it is difficult to anchor the results of the methods, which makes journal evaluation ambiguous and inaccurate.【Method/process】In this paper, based on the principal component analysis method, a new journal evaluation method, the principal component-BP artificial neural network method, is proposed to evaluate 585 comprehensive humanities and social science journals in the Annual Report on Impact Factor of Chinese Academic Journals (Humanities and Social Sciences) (2021), and the evaluation re‐sults are compared with authoritative journal The evaluation results are compared with those of authoritative journals, and then the
evaluation method is analyzed.【Result/conclusion】The research results show that: the principal component-BP artificial neural net‐work method is more accurate than some traditional methods; the principal component-BP artificial neural network method has higher requirements for evaluation objects; meanwhile, To provide some ideas for other fields of journal evaluation as well as evaluation meth‐ods.【Innovation/limitation】This paper only takes humanities and social science journals as an example, and the scope is limited. In the future, we should further expand the scope of research subjects and try to apply this method to evaluation in other fields.