Product evaluation model training method, product evaluation method, device and equipment

[0041] This embodiment provides a training method for a product evaluation model, refer to figure 1 , the training method includes the following steps:

[0042] S1. Obtain sample data of the product, where the sample data includes first sample feature data, second sample feature data, and sample evaluation data;

[0043] S2. Input the first sample feature data into the monotonic network module of the product evaluation model, and input the second sample feature data into the non-monotonic network module of the product evaluation model, wherein the monotonic network module and the non-monotonic network module The network module is combined through a fully connected layer to output the evaluation data of the product; the monotonic relationship between the output value and the input value of the monotonic network module, and the non-monotonic relationship between the output value and the input value of the non-monotonic network module relation;

[0044] S3. Calculate the loss acco

[0053] This embodiment provides a product evaluation method, refer to image 3 , the method includes the following steps:

[0054] S4. Obtain relevant data of the product to be evaluated, where the relevant data includes first characteristic data and second characteristic data;

[0055] S5, inputting the first characteristic data into the monotonic network module of the product evaluation model, inputting the second characteristic data into the non-monotonic network module of the product evaluation model, and outputting the evaluation data of the product to be evaluated;

[0056] Wherein, the product evaluation model is obtained by training according to the training method of Embodiment 1.

[0057] In an implementable solution, the product evaluation method provided in this embodiment further comprises the following steps:

[0058] S6. Sort the multiple products to be evaluated based on the evaluation data of the multiple products to be evaluated.

[0059] refer to Figure 4

[0071] This embodiment provides a training device for a product evaluation model, refer to Image 6 , the device includes:

[0072] A sample acquisition module 1, configured to acquire sample data of a product, the sample data including first sample feature data, second sample feature data and sample evaluation data;

[0073] Classification input module 2, for inputting the first sample feature data into the monotonic network module of the product evaluation model, and inputting the second sample feature data into the non-monotonic network module of the product evaluation model, wherein the monotonic network module It is combined with the non-monotonic network module through the full connection layer to output the evaluation data of the product; the output value of the monotonic network module and the input value are in a monotonic relationship, and the output value of the non-monotonic network module and the input value are in a monotonic relationship. There is a non-monotonic