Form processing method, related device, equipment and storage medium

Abstract

The invention discloses a form processing method, a related device, equipment and a storage medium, and the form processing method comprises the steps: obtaining label relation information of a target form in response to a detected content recommendation instruction for a target form item in the target form; wherein the label relation information at least comprises label elements of form items in the target form and relations among the label elements; and generating target recommendation content of the target form item at least based on the tag relationship information. According to the scheme, the form processing efficiency can be improved.

Claims

1. A form processing method, It is characterized in that include: In response to detecting a content recommendation instruction for a target form item in a target form, obtaining label relationship information of the target form; wherein the label relationship information at least includes label elements of the form items in the target form and the relationship between the label elements; At least based on the tag relationship information, target recommended content of the target form item is generated.

2. The method according to claim 1, It is characterized in that The tag element at least includes a target parameter of a value attribute, and generating target recommended content of the target form item based at least on the tag relationship information includes: Based on the label relationship information, a label element having a target relationship with the label element of the target form item is selected in the target form as a relationship label element; wherein the target relationship includes any one of an adjacent relationship, a same style relationship, and a dependent relationship; The target recommended content is generated based on the target parameter of the relationship tag element with respect to the value attribute.

3. The method according to claim 2, It is characterized in that The step of selecting, in the target form, a label element that has a target relationship with the label element of the target form item based on the label relationship information as a relationship label element includes: Based on the label relationship information, a label element that has the adjacent relationship with the label element of the target form item is selected in the target form as the relationship label element; wherein the adjacent relationship indicates that: the label element of the target form item and the relationship label element have the same number of nested parent label elements, respectively, and the label element of the target form item and the relationship label element have at least one common parent label element.

4. The method according to claim 2, It is characterized in that The label element further includes a target parameter of a style attribute, and based on the label relationship information, selecting a label element in the target form that has a target relationship with the label element of the target form item as a relationship label element in the target form includes: Based on the tag relationship information, a tag element having the same style relationship with the tag element of the target form item is selected in the target form as the relationship tag element; wherein the same style relationship indicates that the tag element of the target form item and the relationship tag element have the same target parameters with respect to the style attributes.

5. The method according to claim 2, It is characterized in that The step of selecting, in the target form, a label element that has a target relationship with the label element of the target form item based on the label relationship information as a relationship label element includes: Based on the label relationship information, a label element having the dependency relationship with the label element of the target form item is selected in the target form as the relationship label element; wherein the dependency relationship indicates that: the label element of the target form item and the relationship label element reference at least one same function.

6. The method according to claim 2, It is characterized in that The label relationship information includes a label relationship graph, the label relationship graph includes a plurality of nodes, the nodes represent label elements of the form items, the target relationship exists between the label elements represented by the connected nodes, and based on the label relationship information, selecting a label element in the target form that has a target relationship with the label element of the target form item as a relationship label element includes: In the label relationship graph, a node connected to a node of a label element representing the target form item is selected as a relationship node; Based on the label relationship graph, a label element represented by the relationship node is selected in the target form as the relationship label element.

7. The method according to claim 6, It is characterized in that The tag element further includes a target parameter of the identifier attribute, and different tag elements have different target parameters for the identifier attribute. The step of acquiring the tag relationship graph includes: constructing the node representing the tag element based on a target parameter of the tag element with respect to the identifier attribute; Whether to connect the nodes is determined based on whether the target relationship exists between the tag elements respectively represented by different nodes.

8. The method according to claim 1, It is characterized in that The generating target recommended content of the target form item based at least on the tag relationship information includes: Based on the tag relationship information, construct a first prompt instruction; wherein the first prompt instruction is used to instruct the large language model to generate recommended content for the target form item based on the tag relationship information; The first prompt instruction is input into the large language model, and output content of the large language model is obtained as the target recommended content.

9. The method according to claim 1, It is characterized in that The tag element includes a target parameter of an identifier attribute, and different tag elements have different target parameters for the identifier attribute. Before generating target recommended content for the target form item based at least on the tag relationship information, the method further includes: Based on the label element of the target form item and the form description information of the target form, a target feature identifier corresponding to the target form item and a feature data table to which the target feature identifier belongs are obtained; wherein the form description information at least includes the subject description information of the target form, and the feature data table at least includes: a feature identifier corresponding to the target parameter of each form item with respect to the identifier attribute, and actual filled-in content of each form item corresponding to the feature identifier; The generating target recommended content of the target form item based at least on the tag relationship information includes: Based on the actual filled-in content corresponding to the target feature identifier in the feature data table, the form description information and the tag relationship information, target recommended content for the target form item is generated.

10. The method according to claim 9, It is characterized in that The step of obtaining a target feature identifier corresponding to the target form item and a feature data table to which the target feature identifier belongs based on the label element and form description information of the target form item includes: Based on the label element of the target form item and the form description information, a second prompt instruction is constructed; wherein the second prompt instruction is used to instruct the large language model to determine the feature identifier corresponding to the target form item and the feature data table to which it belongs by referring to the label element of the target form item and the form description information; The second prompt instruction is input into the large language model, and the output content of the large language model is obtained as the target feature identifier corresponding to the target form item and the feature data table to which it belongs.

11. The method according to claim 9, It is characterized in that The target feature identifier and the feature data table to which the target feature identifier belongs are predicted by a large language model, and the large language model is obtained by fine-tuning parameters based on a sample form. The sample label element of the sample form item in the sample form includes sample parameters of a sample identifier attribute, and different sample label elements have different sample parameters regarding the sample identifier attribute. The sample form is annotated with sample feature information corresponding to each sample form item, and the sample feature information includes the sample feature identifier and the sample feature data table to which the sample feature identifier belongs.

12. The method according to claim 11, It is characterized in that The step of parameter fine-tuning comprises: Based on the sample label element and the sample form description information of the sample target form item, a sample prompt instruction is constructed; wherein the sample form description information at least represents the subject description information about the sample form, and the sample prompt instruction is used to instruct the large language model to predict the predicted feature information corresponding to the sample target form item with reference to the sample label element and the sample form description information of the sample target form item, wherein the predicted feature information includes a predicted feature identifier and a predicted feature data table to which it belongs; The sample prompt instruction is input into the large language model, the output content of the large language model is obtained as the predicted feature information, and based on the difference between the predicted feature information and the sample feature information, the parameters of the large language model are fine-tuned.

13. The method according to claim 1, It is characterized in that After generating the target recommended content of the target form item at least based on the tag relationship information, the method further includes: Fill the target recommended content into the to-be-filled area of ​​the target form item.

14. The method according to claim 1, It is characterized in that Before obtaining the label relationship information of the target form, the method further includes: Obtain the current filled content of the to-be-filled area of ​​the target form item; Based on whether the number of characters in the currently filled-in content exceeds a preset number, it is determined whether to generate the content recommendation instruction.

15. The method according to claim 14, It is characterized in that The content recommendation instruction at least includes a data verification instruction. After generating the target recommended content of the target form item at least based on the tag relationship information, the method further includes: Based on the difference between the target recommended content and the currently filled content of the target form item, determining whether the currently filled content meets the target requirement; In response to the current filled-in content not meeting the target requirement, the current filled-in content is corrected based on the target recommended content, and the corrected current filled-in content is used as new target recommended content.

16. A form processing device, It is characterized in that include: an acquisition module, configured to acquire tag relationship information of the target form in response to detecting a content recommendation instruction for a target form item in a target form; wherein the tag relationship information at least includes tag elements of the form items in the target form and the relationship between the tag elements; A generating module is used to generate target recommended content of the target form item based at least on the tag relationship information.

17. An electronic device, It is characterized in that It comprises a memory and a processor coupled to each other, wherein the processor is used to execute program instructions stored in the memory to implement the form processing method according to any one of claims 1 to 15.

18. A computer-readable storage medium having program instructions stored thereon, It is characterized in that When the program instructions are executed by a processor, the form processing method according to any one of claims 1 to 15 is implemented.

Images