Evaluation Model Construction of Intercultural Adaptability of Chinese-Foreign Joint Education Students Based on Deep Learning

With the rapid expansion of the scale of Sino-foreign cooperation in running schools, the assessment of students' cross-cultural adaptability has become an important link to ensure the quality of education. Traditional evaluation methods rely on fixed weight system and cross-sectional data analysis, which are difficult to capture the dynamic and nonlinear characteristics of acculturation process, resulting in delayed interventions and inaccurate policy implementation. In this study, we propose a dynamic evaluation model based on Long Short-Term Memory (LSTM) network. By constructing a four-dimensional index system covering psychological adaptation, social adaptation, academic adaptation, and environmental adaptation, and combining with grade perception embedding and temporal attention mechanism, the model can realize the adaptive evolution of index weights. The experimental results show that the model achieves a weighted accuracy of $87.3 \%$ in the data of 1,042 students in four school years, which is $9.6 \%$ higher than static model. Its early warning mechanism can identify academic crisis 2.3 semesters in advance, which provides strong support for the quality assurance of education.