关键词: 人文社科, 效率, 可达目标, 联立方程, 神经网络

Abstract: The research on how to improve the input performance of the research of social sciences and humanities is scarce. Based on the provincial panel data of the social sciences and humanities research provided by the Ministry of Education, the paper puts forward a new performance analysis and improvement framework of input factors. Firstly, SBM super efficiency model is used to measure the performance, and the best input with full efficiency is taken as the improvement direction. Then, the performance of input factors is analyzed based on the simultaneous equation model. Finally, BP artificial neural network is used to test the robustness and conduct auxiliary analysis. The results show that the pure technical efficiency is relatively low, the scale efficiency is relatively high and the efficiency has declined in recent years; the R&D fund performance is lower than that of R&D labor force; the contribution of R&D funds improves rapidly in the case of achievable goals, but it is still lower than that of the R&D labor force; there is still a big gap in R&D input performance between the current situation and the situation under achievable goals.

Key words: social sciences and humanities, efficiency, achievable goals, simultaneous equation, neural network