Considering the general and special characteristics of cultural products, identifying the representative indicators of each attribute, we construct a four-dimension attribute assessment system based on “social - economic - technical - ecological” benefits , and propose an interpretable assessment method based on a large model and deep learning for the property value of cross-domain cultural products. Together with the selected four-dimension attribute indicators, the features of different types of cultural products are unified in the assessment index system. Combining the large model and cue word engineering with external key factors to generate quantitative index data, and introducing the attention mechanism to construct the assessment model of property value of cultural products. Dynamic attention is paid to the key features of the assessment results of property value through the large model, and adequate explanations of the assessment results are provided. The model is trained with real transaction data of cultural industry, and experiments of three different assessment methods are compared to verify the model in terms of accuracy, complexity and sensitivity. And the three different types of cultural product assessment outputs demonstrate the effect of the model.