CN116738230A

CN116738230A - Object evaluation model updating method, object evaluation method and device

Info

Publication number: CN116738230A
Application number: CN202310701019.8A
Authority: CN
Inventors: 陈鹏; 李霞; 蔡科
Original assignee: China Construction Bank Corp; CCB Finetech Co Ltd
Current assignee: China Construction Bank Corp; CCB Finetech Co Ltd
Priority date: 2023-06-13
Filing date: 2023-06-13
Publication date: 2023-09-12

Abstract

The disclosure relates to an updating method of an object evaluation model, an object evaluation method and an object evaluation device, and belongs to the technical fields of financial science and technology, information technology industry and artificial intelligence. The method comprises the following steps: acquiring a first data set, inputting the first data set into an object evaluation model, and outputting a first evaluation result set by the object evaluation model; acquiring a second data set updated by the first data set, inputting the second data set into the object evaluation model, and outputting a second evaluation result set by the object evaluation model; judging whether a model updating condition is met or not based on the first evaluation result set and the second evaluation result set; and if the model updating condition is met, updating the object evaluation model. Therefore, the automatic judgment of the model updating condition can be realized, the continuous updating of the model can be realized in the using process of the object evaluation model, the precision of the object evaluation model is improved, and the method is suitable for updating scenes of the credit evaluation model of a user.

Description

Object evaluation model updating method, object evaluation method and device

Technical Field

The present disclosure relates to the technical fields of financial science and technology, information technology industry, and artificial intelligence, and more particularly, to an object evaluation model updating method, an object evaluation device, an electronic apparatus, a computer-readable storage medium, and a computer program product.

Background

At present, with the continuous development of artificial intelligence technology, the model is widely applied in the evaluation field, and has the advantages of high automation degree, low labor cost and the like. For example, the feature data of the object may be input into the object evaluation model, and the evaluation result of the object may be output from the object evaluation model. However, the object evaluation model in the related art has a problem of low accuracy.

Disclosure of Invention

The present disclosure provides an object evaluation model updating method, an object evaluation device, an electronic apparatus, a computer-readable storage medium, and a computer program product, to at least solve the problem of low accuracy of an object evaluation model in the related art. The technical scheme of the present disclosure is as follows:

according to a first aspect of an embodiment of the present disclosure, there is provided a method for updating an object evaluation model, including: acquiring a first data set, inputting the first data set into the object evaluation model, and outputting a first evaluation result set by the object evaluation model; acquiring a second data set updated by the first data set, inputting the second data set into the object evaluation model, and outputting a second evaluation result set by the object evaluation model; judging whether a model updating condition is met or not based on the first evaluation result set and the second evaluation result set; and if the model updating condition is met, updating the object evaluation model.

In one embodiment of the disclosure, the determining whether the model update condition is satisfied based on the first evaluation result set and the second evaluation result set includes: and judging whether the model updating condition is met or not based on the first data set, the second data set, the first evaluation result set and the second evaluation result set.

In one embodiment of the disclosure, the determining whether the model update condition is satisfied based on the first data set, the second data set, the first evaluation result set, and the second evaluation result set includes: obtaining a first index based on the first data set and the second data set; obtaining a second index based on the first evaluation result set and the second evaluation result set; obtaining a third index and a fourth index based on the first evaluation result set; and judging whether the model updating condition is satisfied or not based on at least one of the first index, the second index, the third index and the fourth index.

In one embodiment of the disclosure, the first data set and the second data set include class dimensions to which the object belongs, and the obtaining a first index based on the first data set and the second data set includes: obtaining a first number of objects in each category dimension in the first dataset; acquiring a second number of objects in each category dimension in the second dataset; the first indicator is obtained based on the first number and the second number.

In one embodiment of the disclosure, the obtaining the first indicator based on the first number and the second number includes: obtaining a fifth index in the ith category dimension based on the first quantity and the second quantity in the ith category dimension; obtaining the first index based on the average value of the fifth index under the N category dimensions; wherein i is a positive integer not greater than N, N being a positive integer.

In one embodiment of the disclosure, the first evaluation result set and the second evaluation result set include a rank dimension to which the object belongs, and the obtaining the second index based on the first evaluation result set and the second evaluation result set includes: acquiring a third number of objects in each level dimension in the first evaluation result set; acquiring a fourth number of objects in each level dimension in the second evaluation result set; and obtaining the second index based on the third number and the fourth number.

In one embodiment of the disclosure, the obtaining the second index based on the third number and the fourth number includes: obtaining a sixth index in the ith category dimension based on the variance of the third number in the plurality of class dimensions belonging to the ith category dimension; obtaining a seventh index in the ith category dimension based on the fourth number of variances in the plurality of level dimensions belonging to the ith category dimension; obtaining the second index based on the variance of the sixth index in the N category dimensions and the variance of the seventh index in the N category dimensions; wherein i is a positive integer not greater than N, N being a positive integer.

In one embodiment of the present disclosure, the first evaluation result set includes a rank dimension to which the object belongs, and the obtaining, based on the first evaluation result set, a third index and a fourth index includes: acquiring a third number of objects in each level dimension in the first evaluation result set; obtaining a sixth index in the ith category dimension based on the variance of the third number in the plurality of class dimensions belonging to the ith category dimension; obtaining a third index based on the skewness of the sixth index in the N category dimensions; obtaining a fourth index based on kurtosis of the sixth index under N category dimensions; wherein i is a positive integer not greater than N, N being a positive integer.

In one embodiment of the disclosure, the first data set and the second data set include class dimensions to which the object belongs, and the first evaluation result set includes class dimensions to which the object belongs; wherein the updating the object evaluation model includes: obtaining a fifth index under N category dimensions based on the first data set and the second data set, wherein N is a positive integer; obtaining a sixth index under N category dimensions based on the first evaluation result set; screening out target category dimensions from the N category dimensions based on the fifth index under the N category dimensions and the sixth index under the N category dimensions; determining a first model parameter associated with the target class dimension from model parameters of the object evaluation model; updating the first model parameters.

In one embodiment of the disclosure, the obtaining, based on the first data set and the second data set, a fifth index under N category dimensions includes: obtaining a first number of objects in each category dimension in the first dataset; acquiring a second number of objects in each category dimension in the second dataset; and obtaining a fifth index in the ith category dimension based on the first quantity and the second quantity in the ith category dimension, wherein i is a positive integer not greater than N.

In one embodiment of the disclosure, the obtaining, based on the first evaluation result set, a sixth index under N category dimensions includes: acquiring a third number of objects in each level dimension in the first evaluation result set; and obtaining a sixth index in the ith class dimension based on the variances of the third quantity in the multiple class dimensions belonging to the ith class dimension, wherein i is a positive integer not more than N.

In one embodiment of the present disclosure, the updating the first model parameter includes: obtaining model parameters of at least one set model; and carrying out weighted summation on the first model parameters and at least one model parameter of the set model to obtain second model parameters after the first model parameters are updated.

According to a second aspect of the embodiments of the present disclosure, there is provided an object evaluation method, including: acquiring target data of an object; and inputting the target data into an object evaluation model, and outputting an evaluation result of the object by the object evaluation model, wherein the object evaluation model is obtained by adopting the updating method of the object evaluation model in the first aspect.

According to a third aspect of the embodiments of the present disclosure, there is provided an updating apparatus of an object evaluation model, including: a first acquisition module configured to acquire a first data set, input the first data set into the object evaluation model, and output a first evaluation result set from the object evaluation model; a second acquisition module configured to acquire a second data set updated by the first data set, input the second data set into the object evaluation model, and output a second evaluation result set by the object evaluation model; a judging module configured to judge whether a model update condition is satisfied based on the first evaluation result set and the second evaluation result set; and the updating module is configured to update the object evaluation model if the model updating condition is met.

In one embodiment of the disclosure, the determining module is further configured to: and judging whether the model updating condition is met or not based on the first data set, the second data set, the first evaluation result set and the second evaluation result set.

In one embodiment of the disclosure, the determining module is further configured to: obtaining a first index based on the first data set and the second data set; obtaining a second index based on the first evaluation result set and the second evaluation result set; obtaining a third index and a fourth index based on the first evaluation result set; and judging whether the model updating condition is satisfied or not based on at least one of the first index, the second index, the third index and the fourth index.

In one embodiment of the disclosure, the first data set and the second data set include class dimensions to which the object belongs, and the determining module is further configured to: obtaining a first number of objects in each category dimension in the first dataset; acquiring a second number of objects in each category dimension in the second dataset; the first indicator is obtained based on the first number and the second number.

In one embodiment of the disclosure, the determining module is further configured to: obtaining a fifth index in the ith category dimension based on the first quantity and the second quantity in the ith category dimension; obtaining the first index based on the average value of the fifth index under the N category dimensions; wherein i is a positive integer not greater than N, N being a positive integer.

In one embodiment of the disclosure, the first evaluation result set and the second evaluation result set include a rank dimension to which the object belongs, and the determining module is further configured to: acquiring a third number of objects in each level dimension in the first evaluation result set; acquiring a fourth number of objects in each level dimension in the second evaluation result set; and obtaining the second index based on the third number and the fourth number.

In one embodiment of the disclosure, the determining module is further configured to: obtaining a sixth index in the ith category dimension based on the variance of the third number in the plurality of class dimensions belonging to the ith category dimension; obtaining a seventh index in the ith category dimension based on the fourth number of variances in the plurality of level dimensions belonging to the ith category dimension; obtaining the second index based on the variance of the sixth index in the N category dimensions and the variance of the seventh index in the N category dimensions; wherein i is a positive integer not greater than N, N being a positive integer.

In one embodiment of the disclosure, the first evaluation result set includes a rank dimension to which the object belongs, and the determining module is further configured to: acquiring a third number of objects in each level dimension in the first evaluation result set; obtaining a sixth index in the ith category dimension based on the variance of the third number in the plurality of class dimensions belonging to the ith category dimension; obtaining a third index based on the skewness of the sixth index in the N category dimensions; obtaining a fourth index based on kurtosis of the sixth index under N category dimensions; wherein i is a positive integer not greater than N, N being a positive integer.

In one embodiment of the disclosure, the first data set and the second data set include class dimensions to which the object belongs, and the first evaluation result set includes class dimensions to which the object belongs; wherein the update module is further configured to: obtaining a fifth index under N category dimensions based on the first data set and the second data set, wherein N is a positive integer; obtaining a sixth index under N category dimensions based on the first evaluation result set; screening out target category dimensions from the N category dimensions based on the fifth index under the N category dimensions and the sixth index under the N category dimensions; determining a first model parameter associated with the target class dimension from model parameters of the object evaluation model; updating the first model parameters.

In one embodiment of the present disclosure, the update module is further configured to: obtaining a first number of objects in each category dimension in the first dataset; acquiring a second number of objects in each category dimension in the second dataset; and obtaining a fifth index in the ith category dimension based on the first quantity and the second quantity in the ith category dimension, wherein i is a positive integer not greater than N.

In one embodiment of the present disclosure, the update module is further configured to: acquiring a third number of objects in each level dimension in the first evaluation result set; and obtaining a sixth index in the ith class dimension based on the variances of the third quantity in the multiple class dimensions belonging to the ith class dimension, wherein i is a positive integer not more than N.

In one embodiment of the present disclosure, the update module is further configured to: obtaining model parameters of at least one set model; and carrying out weighted summation on the first model parameters and at least one model parameter of the set model to obtain second model parameters after the first model parameters are updated.

According to a fourth aspect of the embodiments of the present disclosure, there is provided an object evaluation apparatus including: an acquisition module configured to acquire target data of an object; an evaluation module configured to input the target data into an object evaluation model, and output an evaluation result of the object from the object evaluation model, wherein the object evaluation model is obtained by using the update method of the object evaluation model according to the first aspect.

According to a fifth aspect of embodiments of the present disclosure, there is provided an electronic device, comprising: a processor; a memory for storing the processor-executable instructions; wherein the processor is configured to execute the instructions to implement the method of updating an object assessment model as described in the first aspect and/or to implement the method of object assessment as described in the second aspect.

According to a sixth aspect of embodiments of the present disclosure, there is provided a computer readable storage medium, which when executed by a processor of an electronic device, enables the electronic device to perform the method of updating an object evaluation model as described in the previous first aspect and/or to perform the method of object evaluation as described in the previous second aspect.

According to a seventh aspect of embodiments of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the method of updating an object evaluation model as described in the previous first aspect and/or implements the method of object evaluation as described in the previous second aspect.

The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects: the first evaluation result set and the second evaluation result set can be comprehensively considered, whether the model updating condition is met or not is judged, so that automatic judgment of the model updating condition is achieved, when the model updating condition is met, the object evaluation model is updated, so that automatic updating of the object evaluation model is achieved, continuous updating of the model can be achieved in the using process of the object evaluation model, accuracy of the object evaluation model is improved, and the method is suitable for updating scenes of the credit evaluation model of a user.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure and do not constitute an undue limitation on the disclosure.

Fig. 1 is a flowchart illustrating a method of updating an object evaluation model according to a first embodiment of the present disclosure.

Fig. 2 is a flowchart illustrating a method of updating an object evaluation model according to a second embodiment of the present disclosure.

Fig. 3 is a flowchart illustrating a method of updating an object evaluation model according to a third embodiment of the present disclosure.

Fig. 4 is a schematic diagram of an updating method of an object evaluation model according to a fourth embodiment of the present disclosure.

Fig. 5 is a flow chart of an object evaluation method according to a first embodiment of the present disclosure.

Fig. 6 is a block diagram of an updating apparatus of an object evaluation model according to a first embodiment of the present disclosure.

Fig. 7 is a block diagram of an object evaluation apparatus according to a first embodiment of the present disclosure.

Fig. 8 is a block diagram of an electronic device, according to an example embodiment.

Detailed Description

In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions of the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the foregoing figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the disclosure described herein may be capable of operation in sequences other than those illustrated or described herein. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present disclosure as detailed in the accompanying claims.

The data acquisition, storage, use, processing and the like in the technical scheme of the present disclosure all conform to the relevant regulations of the national laws and regulations.

As shown in fig. 1, the method for updating an object evaluation model according to the first embodiment of the present disclosure includes the following steps:

s101, acquiring a first data set, inputting the first data set into a target evaluation model, and outputting a first evaluation result set by the target evaluation model.

The subject of the method for updating the object evaluation model of the present disclosure is an electronic device. The method for updating the object evaluation model according to the embodiment of the present disclosure may be performed by the apparatus for updating the object evaluation model according to the embodiment of the present disclosure, and the apparatus for updating the object evaluation model according to the embodiment of the present disclosure may be configured in any electronic device to perform the method for updating the object evaluation model according to the embodiment of the present disclosure.

It should be noted that the object evaluation model may be any object evaluation model in the related art, and is not limited here too. For example, the object evaluation model may include a credit evaluation model of the user, a performance evaluation model of the commodity, a risk evaluation model of the item, and the like. For example, the object assessment model may be a Light GBM (Light Gradient Boosting Machine, lightweight gradient hoist) model.

In an embodiment of the disclosure, the first data set includes first data of a plurality of users in the user set, and the first evaluation result set includes first evaluation results of the plurality of users in the user set, the first data being in one-to-one correspondence with the first evaluation results. For example, the user set includes users 1 to M, the first data set includes first data of user s, the first evaluation result set includes first evaluation results of user s, s is a positive integer not greater than M, and M is a positive integer.

The first data and the first evaluation result are not limited too much.

For example, taking the object evaluation model as the credit evaluation model of the user as an example, the first data may include characteristic data of the user, and the first evaluation result may include a credit rating, a credit score, and the like of the user. The characteristic data of the user may include, among other things, the user's income, age, gender, industry category, etc.

For example, taking the object evaluation model as a performance evaluation model of the commodity, the first data may include feature data of the commodity, and the first evaluation result may include a performance level, a performance score, a failure rate, and the like of the commodity. Wherein the characteristic data of the commodity may include the material, manufacturing process, appearance, etc. of the commodity.

For example, taking a risk evaluation model in which an object evaluation model is taken as an item as an example, the first data may include feature data of the item, and the first evaluation result may include a risk level, a risk score, and the like of the item. The characteristic data of the project may include the number of people participating in the project, the project period, the investment cost of the project, whether the project implementation area includes a foreign area, and the like.

S102, acquiring a second data set updated by the first data set, inputting the second data set into the object evaluation model, and outputting a second evaluation result set by the object evaluation model.

It should be noted that, the relevant contents of the second data set and the second evaluation result set may refer to the relevant contents of the first data set and the first evaluation result set in the foregoing embodiments, which are not described herein again.

It will be appreciated that there may be inaccuracy in the first data set and that the second data set refers to the updated result of the first data set.

In one embodiment, acquiring the second data set after the first data set update includes determining whether a data set update condition is satisfied, and if the data set update condition is satisfied, acquiring the second data set after the first data set update. Thus, the method can acquire the second data set after the first data set is updated when the data set updating condition is met.

It should be noted that, the data set update condition is not limited too much, for example, the difference between the acquisition time of the first data set and the current time may be greater than or equal to a set threshold, the current time reaches the set time, and so on.

S103, judging whether the model updating condition is met or not based on the first evaluation result set and the second evaluation result set.

In an embodiment of the disclosure, determining whether a model update condition is satisfied based on the first evaluation result set and the second evaluation result set includes the following possible embodiments:

mode 1, based on a first evaluation result set and a second evaluation result set, a second index is obtained, and based on the second index, whether a model update condition is satisfied is determined.

In some examples, determining whether the model update condition is satisfied based on the second indicator includes determining whether the second indicator is greater than or equal to a set threshold, determining that the model update condition is satisfied if the second indicator is greater than or equal to the set threshold, and determining that the model update condition is not satisfied if the second indicator is less than the set threshold.

Mode 2, obtaining a second index based on the first evaluation result set and the second evaluation result set, obtaining a third index and a fourth index based on the first evaluation result set, and judging whether a model updating condition is satisfied based on at least one of the second index, the third index and the fourth index.

In some examples, determining whether the model update condition is met based on at least one of the second indicator, the third indicator, and the fourth indicator includes obtaining a target indicator based on at least one of the second indicator, the third indicator, and the fourth indicator, and determining whether the model update condition is met based on the target indicator.

For example, the calculation process of the target index is as follows:

wherein, sta is a target index, R is a second index, sk is a third index, KT is a fourth index, and b, c and d are coefficients.

For example, based on the target index, whether the model update condition is satisfied is determined, including determining whether the target index is greater than or equal to a set threshold, if the target index is greater than or equal to the set threshold, determining that the model update condition is not satisfied, and if the target index is less than the set threshold, determining that the model update condition is satisfied.

Mode 3, based on the first data set, the second data set, the first evaluation result set, and the second evaluation result set, determining whether a model update condition is satisfied.

Therefore, the method can comprehensively consider the first data set, the second data set, the first evaluation result set and the second evaluation result set to judge whether the model updating condition is met.

In some examples, determining whether the model update condition is met based on the target indicator includes obtaining the target indicator based on the first data set, the second data set, the first evaluation result set, and the second evaluation result set.

In some examples, the target index is obtained based on the first data set, the second data set, the first evaluation result set, and the second evaluation result set, including obtaining multiple types of indexes based on the first data set, the second data set, the first evaluation result set, and the second evaluation result set, and obtaining the target index based on the multiple types of indexes.

It should be noted that, for the content of the second index, the third index and the fourth index, reference may be made to the following embodiments, which are not described herein.

And S104, updating the object evaluation model if the model updating condition is met.

It should be noted that, the model update may be implemented by any model update method in the related art, which is not limited herein.

In one embodiment, updating the object evaluation model includes obtaining a training sample, wherein the training sample includes sample data and sample evaluation results, and updating the object evaluation model based on the training sample.

In one embodiment, updating the object evaluation model includes obtaining model parameters of at least one set model, and weighting and summing the model parameters of the object evaluation model and the model parameters of the at least one set model to obtain updated model parameters of the object evaluation model.

The setting model is not limited too much, and may include, for example, a DNN (Deep Neural Network ) model, an LR (Logistic Regression, logistic regression) model, and the like.

In one embodiment, updating the object evaluation model includes obtaining at least one set model, and combining the object evaluation model and the at least one set model to obtain an updated model of the object evaluation model.

In summary, according to the method for updating the object evaluation model provided by the embodiment of the disclosure, a first data set is acquired, the first data set is input into the object evaluation model, a first evaluation result set is output by the object evaluation model, a second data set updated by the first data set is acquired, the second data set is input into the object evaluation model, a second evaluation result set is output by the object evaluation model, whether a model updating condition is met or not is judged based on the first evaluation result set and the second evaluation result set, and if the model updating condition is met, the object evaluation model is updated. Therefore, whether the model updating condition is met or not can be judged by comprehensively considering the first evaluation result set and the second evaluation result set, so that automatic judgment of the model updating condition is realized, and when the model updating condition is met, the object evaluation model is updated, so that automatic updating of the object evaluation model is realized, namely continuous updating of the model can be realized in the using process of the object evaluation model, the accuracy of the object evaluation model is improved, and the method is suitable for updating scenes of the credit evaluation model of a user.

As shown in fig. 2, the method for updating an object evaluation model according to the second embodiment of the present disclosure includes the following steps:

s201, acquiring a first data set, inputting the first data set into the object evaluation model, and outputting a first evaluation result set by the object evaluation model.

S202, acquiring a second data set updated by the first data set, inputting the second data set into the object evaluation model, and outputting a second evaluation result set by the object evaluation model.

The relevant content of steps S201-S202 can be seen in the above embodiments, and will not be described here again.

S203, obtaining a first index based on the first data set and the second data set.

In the embodiment of the disclosure, the first index is obtained based on the first data set and the second data set, which may include the following possible embodiments:

mode 1 obtains a first index based on data of the same dimension in a first data set and a second data set.

It should be noted that the dimensions are not limited too much, and may include industry category, service life, and the like, for example.

In some examples, the first indicator is derived based on the same-dimensional data in the first data set and the second data set, including deriving the first indicator based on an average of a plurality of differences of the same-dimensional data in the first data set and the second data set.

Mode 2, obtaining a first number of objects in each category dimension in a first data set, obtaining a second number of objects in each category dimension in a second data set, and obtaining a first index based on the first number and the second number.

Thus, the method can comprehensively consider the first number of the objects in each category dimension in the first data set and the second number of the objects in each category dimension in the second data set to obtain the first index.

It should be noted that, the first data set and the second data set include class dimensions to which the object belongs. The category dimensions may include industry categories, gender, merchandise categories, and the like.

It will be appreciated that the first number of different class dimensions may be different or the same, and the second number of different class dimensions may be different or the same, without undue limitation.

In some examples, deriving the first indicator based on the first number and the second number includes deriving a fifth indicator in the ith category dimension based on the first number and the second number in the ith category dimension, and deriving the first indicator based on an average of the fifth indicators in the N category dimensions. Wherein i is a positive integer not greater than N, N being a positive integer. Therefore, according to the method, the fifth index in the category dimension can be obtained based on the first number and the second number in the same category dimension, and the first index can be obtained based on the average value of the fifth index in the plurality of category dimensions.

Note that N is a positive integer not greater than the number of class dimensions, for example, N is the number of class dimensions.

In some examples, deriving the fifth indicator in the ith category dimension based on the first number and the second number in the ith category dimension includes obtaining an absolute difference of the first number and the second number in the ith category dimension, and taking a ratio of the absolute difference to the first number as the fifth indicator in the ith category dimension. For example, the calculation process of the fifth index in the ith category dimension is as follows:

wherein ,C_i X is the fifth index in the ith category dimension _i1 X is the first number in the ith class dimension _i2 A second number in the ith category dimension.

For example, taking the credit rating model with the object rating model as the user as an example, the category dimension is the industry category to which the user belongs, if the first number and the second number of users in the sales category dimension are respectively 100 and 95, the fifth index in the sales category dimension is |100-95|/100=5%, and if the first number and the second number of users in the manufacturing category dimension are respectively 80 and 90, the fifth index in the manufacturing category dimension is |80-90|/80=12.5%.

In some examples, deriving the fifth indicator in the ith category dimension based on the first number and the second number in the ith category dimension includes obtaining an absolute difference between the first number and the second number in the ith category dimension, and taking the absolute difference as the fifth indicator in the ith category dimension.

In some examples, the first indicator is derived based on an average of the fifth indicators in the N category dimensions, including taking the average of the fifth indicators in the N category dimensions as the first indicator. For example, the first index is calculated as follows:

wherein C is a first index, C _i Is the fifth index in the ith category dimension.

In some examples, deriving the first indicator based on the first number and the second number includes inputting the first number and the second number into a setting algorithm, and outputting the first indicator by the setting algorithm.

In some examples, deriving the first indicator based on the first number and the second number includes obtaining a first plurality of variances of the first number and a second plurality of variances of the second number, and deriving the first indicator based on the first and second variances.

In some examples, the first indicator is derived based on the first number and the second number, including obtaining a first average of the first number and a second average of the second number, and the first indicator is derived based on the first average and the second average.

S204, obtaining a second index based on the first evaluation result set and the second evaluation result set.

In the embodiment of the disclosure, the second index is obtained based on the first evaluation result set and the second evaluation result set, which may include the following several possible embodiments:

mode 1, a second index is obtained based on data of the same dimension in the first evaluation result set and the second evaluation result set.

It should be noted that the dimensions are not limited too much, and may include, for example, a credit level, failure rate, risk level, and the like.

In some examples, the second index is derived based on data of the same dimension in the first and second sets of evaluation results, including deriving the second index based on an average of a plurality of differences of the data of the same dimension in the first and second sets of evaluation results.

Mode 2, obtaining a third number of objects in each level dimension in the first evaluation result set, obtaining a fourth number of objects in each level dimension in the second evaluation result set, and obtaining a second index based on the third number and the fourth number.

Thus, the method can comprehensively consider the third number of the objects in each grade dimension in the first evaluation result set and the fourth number of the objects in each grade dimension in the second evaluation result set to obtain the second index.

It should be noted that, the first evaluation result set and the second evaluation result set include a class dimension to which the object belongs. The rank dimension may include a credit rank, a performance rank, a risk rank, and the like.

It will be appreciated that the third number may be different or the same in different level dimensions and the fourth number may be different or the same in different level dimensions, without being excessively limited herein.

In some examples, deriving the second index based on the third number and the fourth number includes deriving a sixth index in the ith class dimension based on variances of the third number in the plurality of class dimensions subordinate to the ith class dimension, deriving a seventh index in the ith class dimension based on variances of the fourth number in the plurality of class dimensions subordinate to the ith class dimension, and deriving the second index based on variances of the sixth index in the N class dimensions and variances of the seventh index in the N class dimensions. Wherein i is a positive integer not greater than N, N being a positive integer. Therefore, the method can obtain the sixth index in the class dimension based on the variances of the third number in the multiple class dimensions belonging to the same class dimension, obtain the seventh index in the class dimension based on the variances of the fourth number in the multiple class dimensions belonging to the same class dimension, and obtain the second index based on the variances of the sixth index and the variances of the seventh index in the multiple class dimensions.

It should be noted that, a membership is provided between a category dimension and a class dimension, and a plurality of class dimensions may belong to the same category dimension.

In some examples, deriving the sixth indicator in the ith class dimension based on the third number of variances in the plurality of rank dimensions that are affiliated with the ith class dimension includes taking the third number of variances in the plurality of rank dimensions that are affiliated with the ith class dimension as the sixth indicator in the ith class dimension.

For example, taking the object evaluation model as the credit evaluation model of the user, the category dimension is the industry category to which the user belongs, the grade dimension is the credit grade to which the user belongs, if the credit grade dimension belonging to the sales industry comprises a first credit grade, a second credit grade and a third credit grade, and the third amounts under the first credit grade, the second credit grade and the third credit grade belonging to the sales industry are respectively Y ₁ 、Y ₂ 、Y ₃ Then the sixth index under the sales industry category dimension is Y ₁ 、Y ₂ 、Y ₃ Is a variance of (c).

In some examples, deriving the seventh indicator in the ith class dimension based on the fourth number of variances in the plurality of rank dimensions that are affiliated with the ith class dimension includes taking the fourth number of variances in the plurality of rank dimensions that are affiliated with the ith class dimension as the seventh indicator in the ith class dimension.

For example, taking the object evaluation model as the credit evaluation model of the user, the category dimension is the industry category to which the user belongs, the grade dimension is the credit grade to which the user belongs, if the credit grade dimension belonging to the sales industry comprises a first credit grade, a second credit grade and a third credit grade, and the fourth amounts under the first credit grade, the second credit grade and the third credit grade belonging to the sales industry are respectively Y ₄ 、Y ₅ 、Y ₆ Then the seventh index under the sales industry category dimension is Y ₄ 、Y ₅ 、Y ₆ Is a variance of (c).

In some examples, the second index is derived based on the variance of the sixth index in the N category dimensions and the variance of the seventh index in the N category dimensions, including deriving the second index based on the difference of the variance of the sixth index in the N category dimensions and the variance of the seventh index in the N category dimensions.

For example, the second index is calculated as follows:

wherein R is a second index, V ₀ Variance of the sixth index under N category dimensions, V ₁ Is the variance of the seventh index in the N category dimensions.

In some examples, deriving the second index based on the third number and the fourth number includes inputting the third number and the fourth number into a setting algorithm, and outputting the second index by the setting algorithm.

In some examples, deriving the second index based on the third number and the fourth number includes obtaining a third variance of the third number and a fourth variance of the fourth number, and deriving the second index based on the third variance and the fourth variance.

In some examples, deriving the second index based on the third number and the fourth number includes obtaining a third average of the plurality of third numbers and a fourth average of the plurality of fourth numbers, and deriving the second index based on the third average and the fourth average.

S205, obtaining a third index and a fourth index based on the first evaluation result set.

In the embodiment of the disclosure, based on the first evaluation result set, the third index and the fourth index are obtained, and the following several possible embodiments may be adopted:

mode 1, obtaining a third number of objects in each class dimension in the first evaluation result set, obtaining a sixth index in the ith class dimension based on variances of the third number in the plurality of class dimensions belonging to the ith class dimension, obtaining a third index based on skewness of the sixth index in the N class dimensions, and obtaining a fourth index based on kurtosis of the sixth index in the N class dimensions. Wherein i is a positive integer not greater than N, N being a positive integer.

Therefore, the method can obtain the sixth index in the class dimension based on the variance of the third number in the multiple class dimensions belonging to the same class dimension, and respectively obtain the third index and the fourth index based on the skewness and kurtosis of the sixth index in the multiple class dimensions.

It should be noted that the first evaluation result set includes a rank dimension to which the object belongs.

In some examples, deriving the third indicator based on the skewness of the sixth indicator in the N category dimensions includes taking the skewness of the sixth indicator in the N category dimensions as the third indicator.

In some examples, the fourth indicator is derived based on kurtosis of the sixth indicator in the N category dimensions, including using kurtosis of the sixth indicator in the N category dimensions as the fourth indicator.

For example, the calculation process of the third index and the fourth index is as follows:

wherein Sk is a third index, KT is a fourth index, and V _0i For a sixth index in the ith category dimension,the average value of the sixth index in the N category dimensions is given, and S is the standard deviation of the sixth index in the N category dimensions.

Mode 2, the first evaluation result set is input into a setting algorithm, and the setting algorithm outputs a third index and a fourth index.

Mode 3, obtaining a third index and a fourth index based on the evaluation results of the same dimension in the first evaluation result set.

In some examples, the third and fourth metrics are derived based on the same dimensional evaluation results in the first set of evaluation results, including deriving the third and fourth metrics based on an average or variance of the same dimensional evaluation results in the first set of evaluation results.

S206, judging whether the model updating condition is met or not based on at least one of the first index, the second index, the third index and the fourth index.

In one embodiment, determining whether the model update condition is satisfied based on at least one of the first index, the second index, the third index, and the fourth index includes determining whether the first index, the second index, the third index, and the fourth index are greater than or equal to respective corresponding set thresholds, determining that the model update condition is satisfied if at least one of the first index, the second index, the third index, and the fourth index is greater than or equal to respective corresponding set thresholds, and determining that the model update condition is not satisfied if the first index, the second index, the third index, and the fourth index are all less than respective corresponding set thresholds.

In one embodiment, determining whether the model update condition is satisfied based on at least one of the first index, the second index, the third index, and the fourth index includes obtaining a target index based on at least one of the first index, the second index, the third index, and the fourth index, and determining whether the model update condition is satisfied based on the target index.

For example, the calculation process of the target index is as follows:

wherein, sta is the target index, C is the first index, R is the second index, sk is the third index, KT is the fourth index, and a, b, C, d is the coefficient.

It should be noted that, based on the target index, whether the relevant content of the model update condition is satisfied is determined, which may be referred to the above embodiment and will not be described herein.

S207, updating the object evaluation model if the model updating condition is satisfied.

The relevant content of step S207 can be seen in the above embodiment, and will not be described here again.

In summary, according to the method for updating the object evaluation model provided by the embodiment of the disclosure, a first index is obtained based on the first data set and the second data set, a second index is obtained based on the first evaluation result set and the second evaluation result set, a third index and a fourth index are obtained based on the first evaluation result set, and whether a model updating condition is satisfied is determined based on at least one of the first index, the second index, the third index and the fourth index. Thus, the first to fourth indexes can be obtained by comprehensively considering the first data set, the second data set, the first evaluation result set and the second evaluation result set, and whether the model updating condition is satisfied or not can be judged by comprehensively considering at least one of the first to fourth indexes.

As shown in fig. 3, the method for updating an object evaluation model according to the third embodiment of the present disclosure includes the steps of:

s301, acquiring a first data set, inputting the first data set into the object evaluation model, and outputting a first evaluation result set by the object evaluation model.

S302, acquiring a second data set updated by the first data set, inputting the second data set into the object evaluation model, and outputting a second evaluation result set by the object evaluation model.

S303, judging whether the model updating condition is met or not based on the first evaluation result set and the second evaluation result set.

The relevant content of steps S301 to S303 can be seen in the above embodiments, and will not be described here again.

And S304, if the model updating condition is met, obtaining a fifth index under N category dimensions based on the first data set and the second data set, wherein N is a positive integer.

In one embodiment, obtaining the fifth index in the N category dimensions based on the first data set and the second data set includes obtaining a first number of objects in each category dimension in the first data set, obtaining a second number of objects in each category dimension in the second data set, and obtaining the fifth index in the i category dimension based on the first number and the second number in the i category dimension, where i is a positive integer not greater than N, to achieve obtaining the fifth index.

The related content of the fifth index may be referred to the above embodiments, and will not be described herein.

And S305, obtaining a sixth index under N category dimensions based on the first evaluation result set.

In one embodiment, obtaining the sixth index in the N class dimensions based on the first set of evaluation results includes obtaining a third number of objects in each class dimension in the first set of evaluation results, and obtaining the sixth index in the i class dimension based on variances of the third number in the plurality of class dimensions belonging to the i class dimension, where i is a positive integer not greater than N, to achieve the obtaining of the sixth index.

It should be noted that, for the related content of the sixth index, reference may be made to the above embodiment, and the description thereof is omitted here.

S306, screening out target category dimensions from the N category dimensions based on the fifth index in the N category dimensions and the sixth index in the N category dimensions.

It should be noted that the number of dimensions of the target class is at least one.

In an embodiment of the present disclosure, based on the fifth index in the N category dimensions and the sixth index in the N category dimensions, selecting the target category dimension from the N category dimensions may include the following possible embodiments:

In the mode 1, the N category dimensions are ranked from large to small according to a fifth index to obtain a first ranking result, the N category dimensions are ranked from large to small according to a sixth index to obtain a second ranking result, the m category dimensions before ranking in the first ranking result are determined to be target category dimensions, and the t category dimensions before ranking in the second ranking result are determined to be target category dimensions. Wherein m and t are positive integers not greater than N.

Thus, in the method, a plurality of category dimensions with a larger fifth index and a plurality of category dimensions with a larger sixth index can be determined as target category dimensions.

Note that neither m nor t is excessively limited, for example, m=t=0.3n. If N is 10, m=t=0.3×10=3.

Mode 2, obtaining an eighth index in the ith category dimension based on the fifth index and the sixth index in the ith category dimension, and screening out a target category dimension from the N category dimensions based on the eighth index in the N category dimensions.

In some examples, obtaining the eighth indicator in the ith category dimension based on the fifth indicator and the sixth indicator in the ith category dimension includes weighted averaging the fifth indicator and the sixth indicator in the ith category dimension to obtain the eighth indicator in the ith category dimension.

In some examples, the target class dimension is selected from the N class dimensions based on an eighth index under the N class dimensions, including sorting the N class dimensions from large to small according to the eighth index to obtain a third sorting result, and determining the class dimension of the p first sorting in the third sorting result as the target class dimension.

S307, determining a first model parameter associated with the dimension of the target category from the model parameters of the object evaluation model.

The model parameters of the object evaluation model and the class dimension have an association relationship, and the class dimension to which the object belongs in the class dimension is obtained based on the model parameters of the object evaluation model associated with the class dimension. For example, taking the credit rating model with the object rating model as the user as an example, the category dimension is the industry category to which the user belongs, the grade dimension is the credit grade to which the user belongs, the credit grade to which the user belongs in the sales industry category dimension is obtained based on the model parameters of the object rating model associated with the sales industry category dimension, and the credit grade to which the user belongs in the manufacturing industry category dimension is obtained based on the model parameters of the object rating model associated with the manufacturing industry category dimension.

It will be appreciated that different class dimensions may be associated with different model parameters or with the same model parameters, without limitation. The first model parameter refers to a model parameter associated with the target class dimension among model parameters of the object evaluation model.

In one embodiment, determining a first model parameter associated with the target class dimension from among the model parameters of the object assessment model includes determining the first model parameter associated with the target class dimension based on an association between the model parameters of the object assessment model and the class dimension. It is understood that the above-mentioned association relationship may be preset.

In one embodiment, the object evaluation model may include sub-models corresponding to class dimensions one by one, and determining a first model parameter associated with the target class dimension from among model parameters of the object evaluation model includes determining the model parameter of the sub-model corresponding to the target class dimension as the first model parameter.

S308, updating the first model parameters.

In one embodiment, updating the first model parameters includes obtaining training samples, wherein the training samples include sample data and sample evaluation results in a target class dimension, and updating the first model parameters based on the training samples.

In one embodiment, updating the first model parameter includes obtaining at least one model parameter of the set model, and weighting and summing the first model parameter and the at least one model parameter of the set model to obtain a second model parameter after the first model parameter is updated. Therefore, the first model parameter and at least one model parameter of the set model can be comprehensively considered, the first model parameter is updated, and the updating precision and efficiency of the model parameter are improved.

It should be noted that, for the first model parameter, the weight of at least one model parameter of the set model is not excessively limited, for example, the set model includes a DNN model and an LR model, and the weight of the first model parameter and the model parameter of the DNN model is greater than or equal to the model parameter of the LR model.

In summary, according to the method for updating the object evaluation model provided by the embodiment of the disclosure, based on the first data set and the second data set, a fifth index under N category dimensions is obtained, based on the first evaluation result set, a sixth index under N category dimensions is obtained, based on the fifth index under N category dimensions and the sixth index under N category dimensions, a target category dimension is screened out from the N category dimensions, a first model parameter associated with the target category dimension is determined from model parameters of the object evaluation model, and the first model parameter is updated. Therefore, the target class dimension can be screened out from the N class dimensions based on the fifth index and the sixth index, only the first model parameters related to the target class dimension are required to be updated, namely, part of model parameters of the object evaluation model are updated, and the updating precision and efficiency of the object evaluation model are improved.

On the basis of any of the above embodiments, as shown in fig. 4, a first data set may be acquired and input into the object evaluation model, a first evaluation result set is output by the object evaluation model, a second data set updated by the first data set is acquired and input into the object evaluation model, and a second evaluation result set is output by the object evaluation model.

Obtaining a first index based on the first data set and the second data set, obtaining a second index based on the first evaluation result set and the second evaluation result set, obtaining a third index and a fourth index based on the first evaluation result set, obtaining a target index Sta based on the first index, the second index, the third index and the fourth index, judging that a model updating condition is not met if the target index Sta is greater than or equal to a set threshold value, judging that the model updating condition is met if the target index Sta is less than the set threshold value, and updating the object evaluation model based on at least one model parameter of the set model.

It should be noted that, based on at least one model parameter of the set model, the relevant content of updating the object evaluation model can be referred to the above embodiment, and will not be described herein.

Fig. 5 is a flowchart illustrating a method of updating an object evaluation model according to a first embodiment of the present disclosure.

As shown in fig. 5, the method for updating the object evaluation model according to the first embodiment of the present disclosure includes the steps of:

s501, target data of an object is acquired.

S502, inputting target data into an object evaluation model, and outputting an object evaluation result by the object evaluation model, wherein the object evaluation model is obtained by adopting an updating method of the object evaluation model.

It should be noted that, the related content of the target data may refer to the related content of the first data in the above embodiment, and the evaluation result of the object and the related content of the object evaluation model may refer to the above embodiment, which is not described herein.

It should be noted that the object evaluation model may be obtained by using the method for updating the object evaluation model described in fig. 1 to 4.

For example, the object evaluation model is taken as a credit evaluation model of the user, the characteristic data of the user may be input into the credit evaluation model of the user, and the credit rating, the credit score, and the like of the user may be output from the credit evaluation model of the user. That is, the target data may include characteristic data of the user, and the evaluation result of the object may include a credit rating, a credit score, and the like of the user.

For example, the object evaluation model is taken as a performance evaluation model of the commodity, the characteristic data of the commodity can be input into the performance evaluation model of the commodity, and the performance grade, performance score, failure rate, and the like of the commodity can be output from the performance evaluation model of the commodity. That is, the target data may include characteristic data of the commodity, and the evaluation result of the object may include performance level, performance score, failure rate, and the like of the commodity.

For example, taking a risk evaluation model in which an object evaluation model is an item as an example, feature data of the item may be input into the risk evaluation model of the item, and a risk level, a risk score, and the like of the item may be output from the risk evaluation model of the item. That is, the target data may include feature data of the item, and the evaluation result of the object may include a risk level, a risk score, and the like of the item.

In summary, according to the object evaluation method provided by the embodiment of the present disclosure, target data of an object is acquired, the target data is input into an object evaluation model, and an evaluation result of the object is output from the object evaluation model, where the object evaluation model is obtained by using an update method of the object evaluation model, and continuous update of the model can be achieved in a use process of the object evaluation model, and the accuracy of the object evaluation model is higher, and further, the accuracy of the object evaluation is higher, so that the object evaluation method is applicable to a scenario of credit evaluation of a user.

As shown in fig. 6, an updating apparatus 600 of an object evaluation model of an embodiment of the present disclosure includes: a first acquisition module 601, a second acquisition module 602, a judgment module 603, and an update module 604.

A first acquisition module 601 configured to acquire a first data set, input the first data set into the object evaluation model, and output a first evaluation result set from the object evaluation model;

a second obtaining module 602, configured to obtain a second data set after the first data set is updated, input the second data set into the object evaluation model, and output a second evaluation result set by the object evaluation model;

a judging module 603 configured to judge whether a model update condition is satisfied based on the first evaluation result set and the second evaluation result set;

an updating module 604 is configured to update the object evaluation model if the model update condition is satisfied.

In one embodiment of the present disclosure, the determining module 603 is further configured to: and judging whether the model updating condition is met or not based on the first data set, the second data set, the first evaluation result set and the second evaluation result set.

In one embodiment of the present disclosure, the determining module 603 is further configured to: obtaining a first index based on the first data set and the second data set; obtaining a second index based on the first evaluation result set and the second evaluation result set; obtaining a third index and a fourth index based on the first evaluation result set; and judging whether the model updating condition is satisfied or not based on at least one of the first index, the second index, the third index and the fourth index.

In one embodiment of the present disclosure, the first data set and the second data set include class dimensions to which the object belongs, and the determining module 603 is further configured to: obtaining a first number of objects in each category dimension in the first dataset; acquiring a second number of objects in each category dimension in the second dataset; the first indicator is obtained based on the first number and the second number.

In one embodiment of the present disclosure, the determining module 603 is further configured to: obtaining a fifth index in the ith category dimension based on the first quantity and the second quantity in the ith category dimension; obtaining the first index based on the average value of the fifth index under the N category dimensions; wherein i is a positive integer not greater than N, N being a positive integer.

In one embodiment of the present disclosure, the first evaluation result set and the second evaluation result set include a rank dimension to which the object belongs, and the determining module 603 is further configured to: acquiring a third number of objects in each level dimension in the first evaluation result set; acquiring a fourth number of objects in each level dimension in the second evaluation result set; and obtaining the second index based on the third number and the fourth number.

In one embodiment of the present disclosure, the determining module 603 is further configured to: obtaining a sixth index in the ith category dimension based on the variance of the third number in the plurality of class dimensions belonging to the ith category dimension; obtaining a seventh index in the ith category dimension based on the fourth number of variances in the plurality of level dimensions belonging to the ith category dimension; obtaining the second index based on the variance of the sixth index in the N category dimensions and the variance of the seventh index in the N category dimensions; wherein i is a positive integer not greater than N, N being a positive integer.

In one embodiment of the present disclosure, the first evaluation result set includes a rank dimension to which the object belongs, and the determining module 603 is further configured to: acquiring a third number of objects in each level dimension in the first evaluation result set; obtaining a sixth index in the ith category dimension based on the variance of the third number in the plurality of class dimensions belonging to the ith category dimension; obtaining a third index based on the skewness of the sixth index in the N category dimensions; obtaining a fourth index based on kurtosis of the sixth index under N category dimensions; wherein i is a positive integer not greater than N, N being a positive integer.

In one embodiment of the disclosure, the first data set and the second data set include class dimensions to which the object belongs, and the first evaluation result set includes class dimensions to which the object belongs; wherein the update module 604 is further configured to: obtaining a fifth index under N category dimensions based on the first data set and the second data set, wherein N is a positive integer; obtaining a sixth index under N category dimensions based on the first evaluation result set; screening out target category dimensions from the N category dimensions based on the fifth index under the N category dimensions and the sixth index under the N category dimensions; determining a first model parameter associated with the target class dimension from model parameters of the object evaluation model; updating the first model parameters.

In one embodiment of the present disclosure, the update module 604 is further configured to: obtaining a first number of objects in each category dimension in the first dataset; acquiring a second number of objects in each category dimension in the second dataset; and obtaining a fifth index in the ith category dimension based on the first quantity and the second quantity in the ith category dimension, wherein i is a positive integer not greater than N.

In one embodiment of the present disclosure, the update module 604 is further configured to: acquiring a third number of objects in each level dimension in the first evaluation result set; and obtaining a sixth index in the ith class dimension based on the variances of the third quantity in the multiple class dimensions belonging to the ith class dimension, wherein i is a positive integer not more than N.

In one embodiment of the present disclosure, the update module 604 is further configured to: obtaining model parameters of at least one set model; and carrying out weighted summation on the first model parameters and at least one model parameter of the set model to obtain second model parameters after the first model parameters are updated.

The specific manner in which the various modules perform the operations in the apparatus of the above embodiments have been described in detail in connection with the embodiments of the method, and will not be described in detail herein.

In summary, the updating device for an object evaluation model provided in the embodiment of the present disclosure obtains a first data set, inputs the first data set into the object evaluation model, outputs a first evaluation result set from the object evaluation model, obtains a second data set updated by the first data set, inputs the second data set into the object evaluation model, outputs a second evaluation result set from the object evaluation model, determines whether a model update condition is satisfied based on the first evaluation result set and the second evaluation result set, and updates the object evaluation model if the model update condition is satisfied. Therefore, whether the model updating condition is met or not can be judged by comprehensively considering the first evaluation result set and the second evaluation result set, so that automatic judgment of the model updating condition is realized, and when the model updating condition is met, the object evaluation model is updated, so that automatic updating of the object evaluation model is realized, namely continuous updating of the model can be realized in the using process of the object evaluation model, the accuracy of the object evaluation model is improved, and the method is suitable for updating scenes of the credit evaluation model of a user.

As shown in fig. 7, an object evaluation device 700 of the embodiment of the present disclosure includes: an acquisition module 701 and an evaluation module 702.

An acquisition module 701 configured to acquire target data of an object;

an evaluation module 702 configured to input the target data into an object evaluation model, and output an evaluation result of the object from the object evaluation model, wherein the object evaluation model is obtained by using the update method of the object evaluation model according to the first aspect.

In summary, the object evaluation device provided in the embodiments of the present disclosure obtains object data of an object, inputs the object data into an object evaluation model, and outputs an object evaluation result from the object evaluation model, where the object evaluation model is obtained by using an update method of the object evaluation model, and can implement continuous update of the model in a use process of the object evaluation model, and the accuracy of the object evaluation model is higher, and further, the accuracy of object evaluation is higher, so that the object evaluation device is suitable for a scenario of credit evaluation of a user.

As shown in fig. 8, the electronic device 800 includes:

memory 810 and processor 820, bus 830 connecting the different components (including memory 810 and processor 820), memory 810 storing a computer program that when executed by processor 820 implements the method of updating and/or method of evaluating an object model according to embodiments of the present disclosure.

Bus 830 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. By way of example, and not limitation, such architectures include Industry Standard Architecture (ISA) bus, micro channel architecture (MAC) bus, enhanced ISA bus, video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

Electronic device 800 typically includes a variety of electronic device readable media. Such media can be any available media that is accessible by electronic device 800 and includes both volatile and nonvolatile media, removable and non-removable media.

Memory 810 may also include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM) 840 and/or cache memory 850. Electronic device 800 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 860 may be used to read from and write to non-removable, non-volatile magnetic media (not shown in FIG. 8, commonly referred to as a "hard disk drive"). Although not shown in fig. 8, a magnetic disk drive for reading from and writing to a removable non-volatile magnetic disk (e.g., a "floppy disk"), and an optical disk drive for reading from or writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In such cases, each drive may be coupled to bus 830 through one or more data medium interfaces. Memory 810 may include at least one program product having a set (e.g., at least one) of program modules configured to carry out the functions of the various embodiments of the disclosure.

A program/utility 880 having a set (at least one) of program modules 870 may be stored, for example, in memory 810, such program modules 870 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each or some combination of which may include an implementation of a network environment. Program modules 870 generally perform the functions and/or methods in the embodiments described in this disclosure.

The electronic device 800 may also communicate with one or more external devices 890 (e.g., keyboard, pointing device, display 891, etc.), one or more devices that enable a user to interact with the electronic device 800, and/or any device (e.g., network card, modem, etc.) that enables the electronic device 800 to communicate with one or more other computing devices. Such communication may occur through an input/output (I/O) interface 892. Also, electronic device 800 may communicate with one or more networks such as a Local Area Network (LAN), a Wide Area Network (WAN) and/or a public network, such as the Internet, through network adapter 893. As shown in fig. 8, network adapter 893 communicates with other modules of electronic device 800 over bus 830. It should be appreciated that although not shown, other hardware and/or software modules may be used in connection with electronic device 800, including, but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, data backup storage systems, and the like.

Processor 820 executes various functional applications and data processing by executing programs stored in memory 810.

It should be noted that, the implementation process and the technical principle of the electronic device in this embodiment refer to the foregoing explanation of the method for updating the object evaluation model and the method for evaluating the object in the embodiments of the present disclosure, which are not repeated here.

In summary, the electronic device provided in the embodiments of the present disclosure may execute the method for updating an object evaluation model and/or the method for evaluating an object as described above, obtain a first data set, input the first data set into the object evaluation model, output a first evaluation result set from the object evaluation model, obtain a second data set updated by the first data set, input the second data set into the object evaluation model, output a second evaluation result set from the object evaluation model, determine whether a model update condition is satisfied based on the first evaluation result set and the second evaluation result set, and update the object evaluation model if the model update condition is satisfied. Therefore, whether the model updating condition is met or not can be judged by comprehensively considering the first evaluation result set and the second evaluation result set, so that automatic judgment of the model updating condition is realized, and when the model updating condition is met, the object evaluation model is updated, so that automatic updating of the object evaluation model is realized, namely continuous updating of the model can be realized in the using process of the object evaluation model, the accuracy of the object evaluation model is improved, and the method is suitable for updating scenes of the credit evaluation model of a user.

To achieve the above embodiments, the present disclosure also proposes a computer-readable storage medium.

Wherein the instructions in the computer-readable storage medium, when executed by the processor of the electronic device, enable the electronic device to perform the method of updating the object evaluation model and/or the method of object evaluation as described above. Alternatively, the computer readable storage medium may be ROM, random Access Memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, etc.

To achieve the above embodiments, the present disclosure further provides a computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the method of updating an object evaluation model and/or the method of object evaluation as described above.

Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This disclosure is intended to cover any adaptations, uses, or adaptations of the disclosure following the general principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It is to be understood that the present disclosure is not limited to the precise arrangements and instrumentalities shown in the drawings, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method of updating an object assessment model, comprising:

acquiring a first data set, inputting the first data set into the object evaluation model, and outputting a first evaluation result set by the object evaluation model;

acquiring a second data set updated by the first data set, inputting the second data set into the object evaluation model, and outputting a second evaluation result set by the object evaluation model;

judging whether a model updating condition is met or not based on the first evaluation result set and the second evaluation result set;

and if the model updating condition is met, updating the object evaluation model.

2. The method of claim 1, wherein the determining whether a model update condition is satisfied based on the first and second sets of evaluation results comprises:

and judging whether the model updating condition is met or not based on the first data set, the second data set, the first evaluation result set and the second evaluation result set.

3. The method of claim 2, wherein the determining whether the model update condition is satisfied based on the first data set, the second data set, the first evaluation result set, and the second evaluation result set comprises:

obtaining a first index based on the first data set and the second data set;

obtaining a second index based on the first evaluation result set and the second evaluation result set;

obtaining a third index and a fourth index based on the first evaluation result set;

and judging whether the model updating condition is satisfied or not based on at least one of the first index, the second index, the third index and the fourth index.

4. A method according to claim 3, wherein the first data set and the second data set comprise class dimensions to which the object belongs, and wherein the deriving the first index based on the first data set and the second data set comprises:

obtaining a first number of objects in each category dimension in the first dataset;

acquiring a second number of objects in each category dimension in the second dataset;

the first indicator is obtained based on the first number and the second number.

5. The method of claim 4, wherein the deriving the first indicator based on the first number and the second number comprises:

obtaining a fifth index in the ith category dimension based on the first quantity and the second quantity in the ith category dimension;

obtaining the first index based on the average value of the fifth index under the N category dimensions; wherein,

i is a positive integer not greater than N, N being a positive integer.

6. The method of claim 3, wherein the first set of evaluation results and the second set of evaluation results comprise a rank dimension to which the object belongs, wherein the deriving the second index based on the first set of evaluation results and the second set of evaluation results comprises:

acquiring a third number of objects in each level dimension in the first evaluation result set;

acquiring a fourth number of objects in each level dimension in the second evaluation result set;

and obtaining the second index based on the third number and the fourth number.

7. The method of claim 6, wherein the deriving the second index based on the third number and the fourth number comprises:

Obtaining a sixth index in the ith category dimension based on the variance of the third number in the plurality of class dimensions belonging to the ith category dimension;

obtaining a seventh index in the ith category dimension based on the fourth number of variances in the plurality of level dimensions belonging to the ith category dimension;

obtaining the second index based on the variance of the sixth index in the N category dimensions and the variance of the seventh index in the N category dimensions; wherein,

i is a positive integer not greater than N, N being a positive integer.

8. A method according to claim 3, wherein the first set of evaluation results includes a rank dimension to which the object belongs, and the deriving third and fourth metrics based on the first set of evaluation results includes:

obtaining a third index based on the skewness of the sixth index in the N category dimensions;

obtaining a fourth index based on kurtosis of the sixth index under N category dimensions; wherein,

i is a positive integer not greater than N, N being a positive integer.

9. The method of any of claims 1-8, wherein the first dataset, the second dataset comprise a category dimension to which an object belongs, and the first set of evaluation results comprise a rank dimension to which the object belongs;

wherein the updating the object evaluation model includes:

obtaining a fifth index under N category dimensions based on the first data set and the second data set, wherein N is a positive integer;

obtaining a sixth index under N category dimensions based on the first evaluation result set;

screening out target category dimensions from the N category dimensions based on the fifth index under the N category dimensions and the sixth index under the N category dimensions;

determining a first model parameter associated with the target class dimension from model parameters of the object evaluation model;

updating the first model parameters.

10. The method of claim 9, wherein the deriving a fifth index in N category dimensions based on the first dataset and the second dataset comprises:

and obtaining a fifth index in the ith category dimension based on the first quantity and the second quantity in the ith category dimension, wherein i is a positive integer not greater than N.

11. The method of claim 9, wherein the deriving a sixth index in N category dimensions based on the first set of evaluation results comprises:

and obtaining a sixth index in the ith class dimension based on the variances of the third quantity in the multiple class dimensions belonging to the ith class dimension, wherein i is a positive integer not more than N.

12. The method of claim 9, wherein updating the first model parameters comprises:

obtaining model parameters of at least one set model;

and carrying out weighted summation on the first model parameters and at least one model parameter of the set model to obtain second model parameters after the first model parameters are updated.

13. An object evaluation method, comprising:

Acquiring target data of an object;

inputting the target data into an object evaluation model, and outputting an evaluation result of the object from the object evaluation model, wherein the object evaluation model is obtained by adopting the updating method of the object evaluation model according to any one of claims 1 to 12.

14. An apparatus for updating an object evaluation model, comprising:

a first acquisition module configured to acquire a first data set, input the first data set into the object evaluation model, and output a first evaluation result set from the object evaluation model;

a second acquisition module configured to acquire a second data set updated by the first data set, input the second data set into the object evaluation model, and output a second evaluation result set by the object evaluation model;

a judging module configured to judge whether a model update condition is satisfied based on the first evaluation result set and the second evaluation result set;

and the updating module is configured to update the object evaluation model if the model updating condition is met.

15. The apparatus of claim 14, wherein the determination module is further configured to:

16. The apparatus of claim 15, wherein the determination module is further configured to:

obtaining a first index based on the first data set and the second data set;

17. The apparatus of claim 16, wherein the first dataset and the second dataset comprise a category dimension to which an object belongs, the determination module further configured to:

18. The apparatus of claim 17, wherein the determination module is further configured to:

i is a positive integer not greater than N, N being a positive integer.

19. The apparatus of claim 16, wherein the first set of evaluation results and the second set of evaluation results comprise a rank dimension to which an object belongs, the determination module further configured to:

and obtaining the second index based on the third number and the fourth number.

20. The apparatus of claim 19, wherein the determination module is further configured to:

i is a positive integer not greater than N, N being a positive integer.

21. The apparatus of claim 16, wherein the first set of evaluation results includes a rank dimension to which the object belongs, the determination module further configured to:

i is a positive integer not greater than N, N being a positive integer.

22. The apparatus of any one of claims 14-21, wherein the first dataset, the second dataset comprise a category dimension to which an object belongs, and the first set of evaluation results comprise a rank dimension to which the object belongs;

Wherein the update module is further configured to:

updating the first model parameters.

23. The apparatus of claim 22, wherein the update module is further configured to:

24. The apparatus of claim 22, wherein the update module is further configured to:

25. The apparatus of claim 22, wherein the update module is further configured to:

obtaining model parameters of at least one set model;

26. An object evaluation device, comprising:

an acquisition module configured to acquire target data of an object;

an evaluation module configured to input the target data into an object evaluation model, and output an evaluation result of the object from the object evaluation model, wherein the object evaluation model is obtained by using the update method of the object evaluation model according to any one of claims 1 to 12.

27. An electronic device, comprising:

a processor;

A memory for storing the processor-executable instructions;

wherein the processor is configured to execute the instructions to implement the method of any of claims 1-13.

28. A computer readable storage medium, which when executed by a processor of an electronic device, causes the electronic device to perform the method of any of claims 1-13.

29. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-13.