Application form generation method based on user credit double-layer fusion model
Patent Information
- Application Number
- CN202510083522.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-20
- Publication Date
- 2025-05-23
AI Technical Summary
In the existing form generation method, the scoring card settings for quantitative evaluation rely on business experience, resulting in unstable credit evaluation, redundant or missing features of the generated application form data, occupying more storage resources or wasting computing resources.
The application form generation method based on the user credit dual-layer fusion model is adopted. Through data validity checksum structured processing, significant difference factor sequences and user profile factors are generated, and the user profile model is combined with the user profile model and the credit grading index model are generated to generate the user credit data set, and then the user application form is generated.
It improves the accuracy of user credit evaluation, reduces the use of storage resources and waste of computing resources, and avoids the problem of redundant data and missing features of application forms.
Smart Images
Figure CN120030015A_ABST
Abstract
Claims
1. A method for generating an application form based on a double-layer fusion model of user credit, comprising: Extracting a user information set from a database, and performing data validity verification on the user information set to obtain a verified user information set, wherein the verified user information set includes verified user information of structured data and unstructured data; Structuring the unstructured verified user information in the verified user information set, and generating a processed user information set in combination with the structured verified user information; Discretizing the processed user information set to generate a significant difference factor sequence, and constructing a significant portrait factor sequence based on the significant difference factor sequence; Generate a user portrait set according to a preset user portrait model, and use each significant portrait factor in the significant portrait factor sequence to perform portrait factor feature aggregation on each user portrait in the user portrait set to generate user portrait score information; Inputting the user portrait set and the user portrait score information into a preset user credit grading index model to generate a user credit data set, wherein the user portrait model and the user credit grading index model are respectively a teacher model and a student model, which are combined into a user credit double-layer fusion model; According to the user credit data set, a user application form is generated for user information data that meets the preset credit conditions for storage.
2. The method according to claim 1, wherein: The method further comprises: In response to receiving a user application form printing operation, determining whether the user credit data associated with the user application form printing operation meets the preset credit condition; In response to the user credit data associated with the user application form printing operation meeting the preset credit condition, controlling the printing device to print the user application form; In response to the user credit data associated with the user application form printing operation not satisfying the preset credit condition, the printing device is controlled to execute an application failure prompt operation.
3. The method according to claim 1, wherein: The performing of data validity verification on the user information set to obtain a verified user information set includes: Performing type verification on the data types included in each user information in the user information set to obtain a data type verification result set; Performing extreme value verification on data values included in each user information in the user information set to obtain a data extreme value verification result set; The user information in the user information set that does not meet the preset validity verification conditions and the corresponding data type verification results and data extreme value verification results are adjusted to generate a verified user information set.
4. The method according to claim 1, wherein: The step of structuring the unstructured verified user information in the verified user information set, and generating a processed user information set in combination with the structured verified user information, includes: For each unstructured verified user information in the verified user information set, perform the following steps: Performing text cleaning on the verified user information to generate cleaned user information; Deduplication is performed on the cleaned text information to obtain deduplicated user information; Filling missing values in the deduplicated user information to generate filled user information; The padded user information is subjected to data structure standardization to generate processed user information.
5. The method according to claim 1, wherein: Each processed user information in the processed user information set includes an index factor and an index value; and The discretization processing of the processed user information set to generate a significant difference factor sequence, and constructing a significant portrait factor sequence according to the significant difference factor sequence, includes: Discretizing the processed user information set to generate a discretized user data set; Performing significance measurement processing on the discretized user data set to generate a significance difference factor sequence, wherein each significance difference factor in the significance difference factor sequence corresponds to a significance measurement value; In response to determining that there is a corresponding significance metric value in the significance difference factor sequence that does not meet the preset model input condition, a secondary significance check is performed on the significance difference factor sequence to generate a significance portrait factor sequence.
6. The method according to claim 5, wherein: Each significant difference factor in the significant difference factor sequence is in the same factor pool; and The generating of a user portrait set according to a preset user portrait model, and performing portrait factor feature aggregation on each user portrait in the user portrait set by using each significant portrait factor in the significant portrait factor sequence to generate user portrait score information, includes: Inputting the processed user information set and each significant portrait factor in the significant portrait factor sequence into the user portrait model to perform a model performance test, so as to generate a user portrait set and user portrait score information, wherein the user portrait model generates the user portrait score information through the following steps; Inputting the backup significant portrait factors in the preset backup factor library and each significant portrait factor in the significant portrait factor sequence into the user portrait model in order of significance to obtain a model performance test value set; Adding the significant difference factors corresponding to the model performance test values less than or equal to the preset test threshold in the model performance test value set to the backup factor library for reuse; Determine the significance difference factor corresponding to the model performance test value greater than the preset test threshold in the model performance test value set as the target difference factor, and add each target difference factor as a significance indicator to a preset indicator library; Each significant indicator in the indicator library and the corresponding significant user distribution portrait are input into the user portrait model to generate a user portrait set and corresponding user portrait score information, wherein the user portrait score information includes the user portrait score value corresponding to each user portrait.
7. The method according to claim 6, wherein: Before inputting the user portrait set and the user portrait score information into a preset user credit rating index model to generate a user credit data set, the method further includes: Determine missing features in the user portrait set using the user portrait score information; Perform feature estimation on each real feature to obtain feature estimation value; Using each feature estimation value, sparse feature repair is performed on each user portrait in the user portrait set to generate a completed user portrait set.
8. An application form generation device based on a double-layer fusion model of user credit, comprising: An acquisition unit is configured to extract a user information set from a database, and perform data validity verification on the user information set to obtain a verified user information set, wherein the verified user information set includes verified user information of structured data and unstructured data; a structured processing unit configured to perform structured processing on the unstructured verified user information in the verified user information set, and to generate a processed user information set in combination with the structured verified user information; a discretization processing unit configured to discretize the processed user information set to generate a significant difference factor sequence, and to construct a significant portrait factor sequence according to the significant difference factor sequence; A significance checking unit is configured to perform significance checking on each significant user distribution portrait in the significant user distribution portrait set according to a preset user portrait model to generate user portrait classification information; An input unit is configured to input the user portrait classification information into a preset user credit grading index model to generate a user credit data set, wherein the user portrait model and the user credit grading index model are a teacher model and a student model respectively, and are combined into a user credit double-layer fusion model; The application form generating unit is configured to generate a user application form for user information data that meets the preset credit conditions according to the user credit data set, so as to store the form.
9. An electronic device, comprising: one or more processors; a storage device having one or more programs stored thereon, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7.
10. A computer readable medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.