CN118484337A - Workflow recovery method and device - Google Patents

Abstract

The present invention discloses a method and device for workflow recovery, and relates to the field of computer technology. A specific implementation of the workflow recovery method includes: in response to obtaining the termination information of the workflow, generating a corresponding prompt word according to the termination information; inputting the prompt word into a pre-set large model, so that the large model generates guidance information for the workflow according to the prompt word; displaying the guidance information to a pre-set front-end page, so that the user who processes the workflow can return to the process node included in the workflow according to the guidance information and continue to process the workflow. This implementation parses the termination information of the workflow based on the large model, obtains the corresponding guidance information, and enables the user to return to the process node of the workflow according to the guidance information and continue to process the workflow, which can improve the processing efficiency of the workflow, reduce the difficulty of users processing the workflow, and improve the user experience.

Description

A method and device for workflow recovery

Technical Field

The present invention relates to the field of computer technology, and in particular to a method and device for recovering a workflow.

Background Art

In business systems, workflow business scenarios are very common, and users often need to process workflows. For example, before purchasing insurance services, users need to process workflows including: entering basic information, asking relevant questions, and applying for quotes and underwriting from insurance companies. Workflows usually consist of multiple pages, and users need to submit information and complete necessary interactive operations on the page according to the page prompts, such as clicking, dragging, checking, etc., and gradually process the workflow corresponding to each page until the entire workflow is completed. In the process of processing the workflow, if the user has questions or a page anomaly occurs, they can self-diagnose through the page prompts, or consult traditional robot customer service for processing suggestions.

In the process of implementing the present invention, the inventors found that the prior art has at least the following problems:

The page prompts and the terms used in traditional robot customer service are often more professional and have a certain usage threshold. When users have questions or page anomalies occur, they cannot provide users with accurate help, resulting in low workflow processing efficiency.

Summary of the invention

In view of this, an embodiment of the present invention provides a method and apparatus for recovering a workflow, which can improve the processing efficiency of the workflow, reduce the difficulty of users processing the workflow, and improve the user experience.

To achieve the above-mentioned purpose, according to the first aspect of an embodiment of the present invention, a method for workflow recovery is provided, comprising: in response to obtaining termination information of the workflow, generating a corresponding prompt word according to the termination information; inputting the prompt word into a pre-set large model, so that the large model generates guidance information of the workflow according to the prompt word; displaying the guidance information on a pre-set front-end page, so that the user processing the workflow can return to the process node included in the workflow according to the guidance information and continue processing the workflow.

Optionally, a corresponding prompt word is generated based on the termination information, including: determining the termination scenario of the workflow based on the termination information; determining the guide scenario corresponding to the termination scenario based on pre-set business orchestration information; updating the prompt word template corresponding to the guide scenario based on the termination information, and using the updated prompt word template as the prompt word.

Optionally, determining a guiding scenario corresponding to the termination scenario according to pre-set business orchestration information includes: determining a business logic corresponding to the termination scenario according to the business orchestration information; executing the business logic to obtain an execution result of the business logic; and determining a guiding scenario corresponding to the termination scenario according to the execution result.

Optionally, before displaying the guidance information on a preset front-end page, the intention information associated with the guidance information is determined, and it is determined that the intention information does not meet the preset guidance conditions; if the intention information meets the preset guidance conditions, the following steps are executed in a loop to determine the guidance information to be displayed on the front-end page:

Use the intent information as the intent information of the first round, determine the user purpose information corresponding to the intent information of the current round, and generate new prompt words for the current round according to the user purpose information; determine the big model of the current round according to the big model, input the new prompt words of the current round into the big model of the current round, and obtain the new intent information and new guidance information output by the big model of the current round; when the new intent information of the current round does not meet the guidance conditions, the loop is terminated, and the new guidance information of the current round is used as the guidance information displayed on the front-end page; when the new intent information of the current round meets the guidance conditions, the new intent information of the current round is used as the intent information for the next round.

Optionally, determining the big model of the current round based on the big model includes: in response to determining that the new prompt word of the current round matches the big model, using the big model as the big model of the current round; in response to determining that the new prompt word does not match the big model, screening out a target big model that matches the new prompt word of the current round from a plurality of pre-set big models; and using the target big model as the big model of the current round.

Optionally, before obtaining the new guidance information and new intent information output by the big model of the current round, historical guidance information associated with the user is obtained; the historical guidance information is input into the big model of the current round, so that the big model of the current round outputs the new guidance information and new intent information of the current round according to the new prompt words of the current round and the historical guidance information.

Optionally, obtaining the termination information of the workflow includes: in response to receiving a dialogue request, determining the user's dialogue information based on the dialogue request, and using the dialogue information as the termination information of the workflow; or, in response to detecting the occurrence of a target event, determining whether the target event belongs to a pre-set termination event set, and if the target event belongs to the termination event set, determining event information of the target event, and using the event information as the termination information of the workflow.

According to a second aspect of an embodiment of the present invention, there is provided a device for workflow recovery, comprising:

A first generating module, configured to generate corresponding prompt words according to the obtained workflow termination information;

The second generation module is used to input the prompt word into a preset large model, so that the large model generates workflow guidance information according to the prompt word;

The guidance module is used to display the guidance information on a preset front-end page, so that the user processing the workflow can return to the process node included in the workflow according to the guidance information and continue processing the workflow.

Optionally, a corresponding prompt word is generated based on the termination information, including: determining the termination scenario of the workflow based on the termination information; determining the guide scenario corresponding to the termination scenario based on pre-set business orchestration information; updating the prompt word template corresponding to the guide scenario based on the termination information, and using the updated prompt word template as the prompt word.

Optionally, determining a guiding scenario corresponding to the termination scenario according to pre-set business orchestration information includes: determining a business logic corresponding to the termination scenario according to the business orchestration information; executing the business logic to obtain an execution result of the business logic; and determining a guiding scenario corresponding to the termination scenario according to the execution result.

Optionally, the guidance module is further used to: before displaying the guidance information on a preset front-end page, determine the intention information associated with the guidance information, and determine that the intention information does not meet the preset guidance conditions;

When the intention information meets the preset guidance conditions, the following steps are executed in a loop to determine the guidance information to be displayed on the front-end page:

Use the intent information as the intent information of the first round, determine the user purpose information corresponding to the intent information of the current round, and generate new prompt words for the current round according to the user purpose information; determine the big model of the current round according to the big model, input the new prompt words of the current round into the big model of the current round, and obtain the new intent information and new guidance information output by the big model of the current round; when the new intent information of the current round does not meet the guidance conditions, the loop is terminated, and the new guidance information of the current round is used as the guidance information displayed on the front-end page; when the new intent information of the current round meets the guidance conditions, the new intent information of the current round is used as the intent information for the next round.

Optionally, the guidance module is also used to: in response to determining that the new prompt word of the current round matches the big model, use the big model as the big model of the current round; in response to determining that the new prompt word does not match the big model, screen out a target big model that matches the new prompt word of the current round from a plurality of pre-set big models; and use the target big model as the big model of the current round.

Optionally, the guidance module is also used to: obtain historical guidance information associated with the user; input the historical guidance information into the big model of the current round, so that the big model of the current round outputs new guidance information and new intention information of the current round according to the new prompt words of the current round and the historical guidance information.

Optionally, obtaining the termination information of the workflow includes: in response to receiving a dialogue request, determining the user's dialogue information based on the dialogue request, and using the dialogue information as the termination information of the workflow; or, in response to detecting the occurrence of a target event, determining whether the target event belongs to a pre-set termination event set, and if the target event belongs to the termination event set, determining event information of the target event, and using the event information as the termination information of the workflow.

According to a third aspect of an embodiment of the present invention, there is provided an electronic device, including:

one or more processors;

a storage device for storing one or more programs,

When the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of the above embodiments.

According to a fourth aspect of an embodiment of the present invention, there is provided a computer-readable medium on which a computer program is stored, and when the program is executed by a processor, the method of any of the above embodiments is implemented.

According to a fifth aspect of an embodiment of the present invention, there is provided a computer program product, comprising a computer program, which implements the method of any of the above embodiments when executed by a processor.

One embodiment of the above invention has the following advantages or beneficial effects: based on the big model, the termination information of the workflow is parsed to obtain the corresponding guidance information, so that the user can return to the process node of the workflow according to the guidance information and continue to process the workflow, which can improve the processing efficiency of the workflow, reduce the difficulty of users processing the workflow, and improve the user experience; according to the business arrangement information, the guidance scenario corresponding to the termination scenario is determined, and prompt words are generated according to the guidance scenario, and prompt words can be generated flexibly and accurately; the business logic corresponding to the termination scenario is executed, and the guidance scenario is determined according to the execution result of the business logic, and the guidance scenario can be determined flexibly and efficiently; according to the intention information associated with the guidance information, it is judged whether it is necessary to generate new guidance information and intention information again, and after multiple big model outputs, the guidance scenario can be obtained. The guidance information finally returned to the user can accurately determine the user's intention and purpose, facilitate the provision of accurate guidance information to the user, and help the user quickly return to the process node of the workflow; when the prompt word does not match the large model, the target large model is screened out from multiple large models, which can improve the flexibility of the large model output and meet different large model analysis needs; based on the historical guidance information previously output to the user, new guidance information is generated, which can further improve the accuracy of the guidance information and improve the workflow processing efficiency; the triggering timing of the method includes: the user asks a question to the execution subject of the embodiment of the present invention or requests a dialogue, and detects that the workflow is abnormally exited or abnormally interrupted due to certain operations of the user, which can reduce the user's usage difficulty and improve the user's efficiency in processing the workflow.

The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner.

BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

FIG1 is a schematic diagram of the main process of a method for workflow recovery according to an embodiment of the present invention;

FIG2 is a schematic diagram of an application scenario of workflow recovery according to a reference embodiment of the present invention;

FIG3 is a schematic diagram of a large model dialogue chain according to a reference embodiment of the present invention;

FIG4 is a schematic diagram of a technical architecture of workflow recovery according to a reference embodiment of the present invention;

FIG5 is a schematic diagram of the main process of workflow recovery in multi-module collaboration according to a reference embodiment of the present invention;

FIG6 is a schematic diagram of the main process of a method for workflow recovery according to a reference embodiment of the present invention;

7 is a schematic diagram of main modules of a device for workflow recovery according to an embodiment of the present invention;

FIG8 is a diagram of an exemplary system architecture in which an embodiment of the present invention may be applied;

FIG. 9 is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention.

DETAILED DESCRIPTION

The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

It should be noted that in the technical solution of the present invention, the collection, use, storage, sharing and transfer of user personal information involved are in compliance with the provisions of relevant laws and regulations, and it is necessary to inform the user and obtain the user's consent or authorization. When applicable, the user's personal information is de-identified and/or anonymized and/or encrypted.

In business systems, workflow business scenarios are very common, and users often need to process workflows. For example, before purchasing insurance services, users need to process workflows including: entering basic information, asking relevant questions, and applying for quotes and underwriting from insurance companies. Workflows usually consist of multiple pages, and users need to submit information and complete necessary interactive operations on the page according to the page prompts, such as clicking, dragging, checking, etc., and gradually process the workflow corresponding to each page until the entire workflow is completed. In the process of processing the workflow, if the user has questions or a page anomaly occurs, they can self-diagnose through the page prompts, or consult traditional robot customer service for processing suggestions.

The page prompts and the terms used in traditional robot customer service are often more professional and have a certain usage threshold. When users have questions or page anomalies occur, they cannot provide users with accurate help, resulting in low workflow processing efficiency and poor user experience.

In view of this, according to a first aspect of an embodiment of the present invention, a method for recovering a workflow is provided.

Fig. 1 is a schematic diagram of the main process of the method for workflow recovery according to an embodiment of the present invention. As shown in Fig. 1, the method for workflow recovery according to an embodiment of the present invention mainly includes the following steps S101 to S103.

Step S101, in response to obtaining the termination information of the workflow, generating a corresponding prompt word according to the termination information.

The execution subject of the embodiment of the present invention monitors the process of users processing workflows, obtains user behavior data, form collection data, page processing results and other workflow-related information generated by users in the process of processing workflows, and extracts termination information from the above information.

Exemplarily, the workflow processed by the user includes multiple pages. The execution subject of the embodiment of the present invention obtains the page processing result of each page, and the page processing result includes: whether the processing is successful, the information entered by the user on the page form, the basic information displayed on the page, etc. If the page processing result is "successful", the workflow of the next page is displayed to the user. If the page processing result is "unsuccessful", the user's workflow processing flow is determined to be terminated, and the user behavior data such as the text entered by the user on the page and the uploaded file, the number of times the user clicks the button, the number of times the user refreshes the page, and the basic information displayed on the page are obtained, and the above-mentioned information is used as the termination information.

After the suspension information is acquired, the suspension information is filled into a preset prompt word template, and the prompt word template after the suspension information is filled in is used as the prompt word. Exemplarily, the preset prompt word template is "You are a customer service staff. A problem occurred when the user was processing the workflow, resulting in the workflow being terminated. The relevant information includes: the text entered by the user on the page: [A1], the user's behavior data: [A2], the basic information displayed on the page: [A3], the workflow processing result: [A4], please provide a solution for the user based on the above information and guide the user to complete the workflow processing", and the corresponding termination information replaces A1, A2, A3 and A4 in the above prompt word template. The prompt word template after replacement becomes "You are a customer service staff. A problem occurred when the user was processing the workflow, resulting in the workflow being terminated. The relevant information includes: the text entered by the user on the page: user, 20,..., the user's behavior data: contact clicked the "Confirm" button 5 times, the basic information displayed on the page: name, age,..., the workflow processing result: unsuccessful, some information processing was abnormal, please provide a solution for the user based on the above information and guide the user to complete the workflow processing".

It should be noted that the prompt word template is configurable, and the execution subject of the embodiment of the present invention adds a new prompt word template, or modifies or deletes an existing prompt word template according to a received prompt word update request.

By monitoring the workflow processing process, the workflow termination information can be obtained in time. According to the prompt word template, the termination information can be converted into prompt words, which can provide a data basis for the large model, improve the workflow processing efficiency, and enable users to return to the workflow processing process in time.

Step S102: input the prompt word into a preset macro model, so that the macro model generates workflow guidance information according to the prompt word.

After the prompt words are generated, the prompt words are input into the pre-set big model, which is a generative big model with business knowledge of the business field to which the workflow belongs. Specifically, before using the big model to output guidance information, the big model is pre-trained using business knowledge so that the big model is familiar with various processes in the business field and has existing business knowledge in the business field. Then the prompt words are input into the pre-trained big model.

The large model outputs corresponding guidance information based on the received prompt words. The guidance information is used to help users continue to process the previously terminated workflow. The guidance information includes: a description of the current workflow processing status, the reason for the workflow termination, and the operations that the user can perform (i.e., the method of returning to the previously terminated workflow process node).

For example, the guidance information output by the large model is: the process node where the workflow was just terminated is node B1, the meaning of the process node is data B2, the information you need to provide is information B3, if you have no other questions, you can click page link B4 to continue the process node that was just terminated. It should be noted that in different guidance information, the values of node B1, data B2, information B3 and page link B4 are different, and are generated according to the received termination information.

This guidance information can answer questions that users have when processing workflows, allowing users to more clearly understand the workflow processing process, and enable users to return to previously terminated workflows in a timely manner through page links, helping users improve workflow processing efficiency and accuracy and enhance user experience.

Step S103, displaying the guidance information on a preset front-end page, so that the user processing the workflow can return to the process node included in the workflow according to the guidance information and continue processing the workflow.

After the guidance information is generated, the guidance information is displayed on the front page. Specifically, the front page is a pre-set dialog box, and the execution subject of the embodiment of the present invention communicates with the user as a page intelligent robot, and the dialogue content is the previously generated guidance information.

Based on the guidance information, the user understands the meaning of the workflow, knows the information that needs to be filled in and the operations that need to be performed, and then clicks the link in the guidance information to send a process node access request to the execution subject of the embodiment of the present invention. The execution subject of the embodiment of the present invention parses the received process node access request, determines the target process node that the user wants to access, and displays the page corresponding to the target process node to the user, which is equivalent to the user returning to the previously terminated workflow, and the user continues to process the previously unfinished workflow in the page corresponding to the target process node.

It should be noted that after receiving the guidance information, the user can continue to communicate with the page intelligent robot (i.e., the execution subject of the embodiment of the present invention) in the dialog box, for example, input text, images, or videos and other information in the dialog box, and the execution subject of the embodiment of the present invention will generate new guidance information based on the user's dialogue content and display it in the dialog box. Specifically, the user's dialogue content is converted into a new prompt word and input into the large model to obtain new guidance information. That is, the above steps S101 to S103 are repeated in the form of a dialogue between the page intelligent robot and the user. The user sends the dialogue content to the page intelligent robot, and the page intelligent robot uses the user's dialogue content as the termination information, generates corresponding guidance information based on the large model, and displays the guidance information as the reply content in the dialog box, and repeats the above steps until the user returns to the previously terminated workflow according to the link in the guidance information.

FIG2 is a schematic diagram of an application scenario of workflow recovery according to a reference embodiment of the present invention. Exemplarily, as shown in FIG2, the conversation content between the user (i.e., user) and the execution subject (i.e., AI) of the embodiment of the present invention is displayed in the conversation page 201. When the execution subject of the embodiment of the present invention detects that the workflow processed by the user is suspended, it collects suspension information, determines that the workflow is suspended in the "product usage description link", generates the first guidance information (i.e., the first sentence sent by AI) based on the big model, and displays the guidance information in the conversation page 201 to have a conversation with the user; the user explains his doubts (i.e., the first sentence sent by the user) based on the first guidance information, and the execution subject of the embodiment of the present invention generates the second guidance information (i.e., the second sentence sent by AI) based on the big model again according to the doubts sent by the user. The second guidance information includes a page link, which can help the user return to the previously suspended workflow. If the user wants to continue communicating, he can enter the content in the dialog box at the bottom of the conversation page 201 and send it to the execution subject of the embodiment of the present invention. The execution subject of the embodiment of the present invention will continue to output guidance information based on the content sent by the user to help the user answer questions and solve doubts.

Based on the big model, the workflow termination information is parsed to obtain the corresponding guidance information, so that the user can return to the process node of the workflow according to the guidance information and continue to process the workflow. This can improve the processing efficiency of the workflow, reduce the difficulty of users processing the workflow, and improve the user experience.

According to a reference embodiment of the present invention, when generating corresponding prompt words according to the termination information, first determine the termination scenario of the workflow according to the termination information. Specifically, the termination scenario includes: the page being visited, whether the page is closed, whether there is user input in the page, whether the workflow corresponding to the page is processed successfully, and other page states. The termination scenario is a collection of the above page states. According to the termination information, the above page states are determined, and the page states are combined into a termination scenario. For example, when the termination information includes: the user visited page C, page C was closed, and the workflow corresponding to page C was not processed successfully, the above termination information is converted into a termination scenario, and "page C is closed and there is no user input" is obtained.

After determining the suspension scenario, determine the guidance scenario corresponding to the suspension scenario according to the preset business arrangement information. The business arrangement information includes the association between the suspension scenario and the guidance scenario. Each suspension scenario corresponds to a guidance scenario. For example, when the suspension scenario is "page C1 is closed and there is no user input", query the business arrangement information to determine that the corresponding guidance scenario is "asking the user whether to access page C2".

For each change that occurs on the front-end page and causes the workflow to be terminated (for example, the user completes the page process, the user asks a question, the user performs an abnormal operation, etc.), a corresponding termination scenario and a guidance scenario corresponding to the termination scenario are pre-set in the business arrangement information. It should be noted that the business arrangement information is configurable, and the execution subject of the embodiment of the present invention updates the business arrangement information according to the received update request, adds a new termination scenario and/or guidance scenario, and modifies the association relationship between the termination scenario and the guidance scenario.

Each guidance scene corresponds to a prompt word template, and the prompt word template is an incomplete prompt word. After the guidance scene is determined, the prompt word template corresponding to the guidance scene is updated according to the suspension information. Specifically, the suspension information is filled into the prompt word template. The prompt word template after the suspension information is filled (i.e., updated) is a complete prompt word. The updated prompt word template is used as the prompt word and input into the pre-set large model to obtain the corresponding guidance information.

Determine the abort scenario based on the abort information, determine the guidance scenario associated with the abort scenario based on the business arrangement information, and then determine the prompt word template based on the guidance scenario, so that the prompt word template can be determined more flexibly and accurately, and it is convenient to generate accurate guidance information. The business arrangement information is configurable, and the abort scenario can be more flexibly associated with the guidance scenario to meet different business processing requirements and efficiently and accurately return the user to the workflow.

According to another reference embodiment of the present invention, when determining the guidance scenario corresponding to the suspension scenario according to the preset business arrangement information, the business logic corresponding to the suspension scenario is first determined according to the business arrangement information. Specifically, the business arrangement information is used to manage the business logic under different suspension scenarios, and the business arrangement information includes the business logic corresponding to each suspension scenario, and the business logic guides the user to return to the workflow and continue to process the workflow.

Execute the business logic corresponding to the abort scenario and obtain the execution result of the business logic. The data type of the execution result can be a string, such as "yes", "no", "success", "unsuccessful", or a numeric type, such as "10", "20", or a Boolean type, such as "true", "false". After obtaining the execution result, determine the guidance scenario corresponding to the abort scenario based on the execution result. For example, when the data type of the execution result is a numeric type, compare the execution result with multiple pre-set numerical ranges, each numerical range corresponds to a guidance scenario, and use the numerical range to which the execution result belongs as the target numerical range, and use the guidance scenario corresponding to the target numerical range as the guidance scenario corresponding to the abort scenario.

Exemplarily, the termination scenario is "an exception occurs on page D", and the business logic corresponding to the termination scenario is the method "http://api/test". The above business logic is called to obtain the execution result of the business logic. When the execution result is "yes", the guidance scenario of the termination scenario is determined to be the guidance scenario E1. When the execution result is "no", the guidance scenario of the termination scenario is determined to be the guidance scenario E2.

It should be noted that the business logic corresponding to the suspension scenario can be implemented by means of a business interface or a custom script. The business logic corresponding to the suspension scenario is configurable. The execution subject of the embodiment of the present invention modifies the business interface or custom script corresponding to the suspension scenario according to the received update request, and then modifies the business logic corresponding to the suspension scenario.

The business logic corresponding to the suspension scenario is executed, and the guidance scenario corresponding to the suspension scenario is determined according to the execution result of the business logic. The guidance scenario can be determined flexibly and accurately, thereby improving the accuracy of the guidance information.

According to another reference embodiment of the present invention, before displaying the guidance information to the preset front-end page, the intention information associated with the guidance information is determined, and it is determined that the intention information does not meet the preset guidance conditions. While the large model generates guidance information based on the prompt word, it also outputs the corresponding intention information, which is used to describe the user's intention, including: purchase, consultation, appointment, return to the previous page, etc.

The guidance conditions include an intention information set. When the intention information output by the large model belongs to the intention information set, it is determined that the intention information meets the guidance conditions, indicating that the guidance information output by the large model is inaccurate and further processing is required to obtain more accurate guidance information.

Specifically, when the intent information meets the guidance conditions, the following steps are executed in a loop to determine the guidance information displayed to the front-end page: the intent information is used as the intent information of the first round, the user purpose information corresponding to the intent information of the current round is determined, and the new prompt words of the current round are generated according to the user purpose information; the large model of the current round is determined according to the large model, and the new prompt words of the current round are input into the large model of the current round to obtain the new intent information and new guidance information output by the large model of the current round; when the new intent information of the current round does not meet the guidance conditions, the loop is terminated, and the new guidance information of the current round is used as the guidance information displayed to the front-end page; when the new intent information of the current round meets the guidance conditions, the new intent information of the current round is used as the intent information of the next round. Specifically, the pre-set business arrangement information is queried to determine the purpose information associated with the intent information. For example, when the intent information is "purchase", the corresponding purpose information is "obtain purchase discount consultation". According to the user purpose information, the corresponding prompt word template is filled, the corresponding new prompt word is generated, the new prompt word is input into the large model, and the new guidance information and new intent information output by the large model are obtained. Then determine whether the new intention information meets the guidance conditions.

FIG3 is a schematic diagram of a large model dialogue chain according to a reference embodiment of the present invention. Exemplarily, as shown in FIG3, the user initiates a dialogue on the front-end page, triggering the workflow termination mechanism. The execution subject of the embodiment of the present invention uses the text entered by the user on the front-end page as the termination information, and generates prompt word 1 according to the termination information; then the prompt word 1 is input into the large model to generate guide information 1 and intention information 1; it is determined whether the intention information 1 is the target intention, and if the intention information 1 is not the target intention, the guide information 1 is displayed on the front-end page and returned to the user; if the intention information 1 is the target intention, prompt word 2 is generated according to the termination information and the guide information 1; then prompt word 2 is input into the large model to generate guide information 2 and intention information 2; then the previous judgment steps are repeated until the intention information is not the target intention, and the corresponding guide information is returned to the user. It should be noted that the input for generating the guide information includes: the text entered by the user on the front-end page, the user behavior data, and the guide information generated and returned to the user before; the generation process of the guide information is regarded as an input and output process, and the guide information output by the previous process is the information input by the next process.

Based on the intent information associated with the guidance information, it is determined whether new guidance information and intent information need to be generated again. After multiple large model outputs, the guidance information finally returned to the user is obtained, which can accurately determine the user's intention and purpose, facilitate the provision of accurate guidance information to the user, and help the user quickly return to the process node of the workflow.

According to a reference embodiment of the present invention, when determining the big model of the current round according to the big model, it is determined whether the new prompt word of the current round matches the big model. For example, the prompt word template corresponding to the new prompt word is determined, and the pre-set business arrangement information is queried to determine whether the big model associated with the prompt word template is the big model to be input. If so, it is determined that the new prompt word matches the big model. For another example, the pre-set business arrangement information is queried to determine the prompt word conditions corresponding to the big model. The prompt word conditions include: the maximum length of the prompt word, the keywords that the prompt word should (or should not) include, the prompt word does not have a null value, etc. When the new prompt word meets the prompt word conditions corresponding to the big model, it is determined that the new prompt word matches the big model. In response to determining that the new prompt word of the current round matches the big model, the big model is used as the big model of the current round, and the new prompt word is input into the big model.

In response to determining that the new prompt word of the current round does not match the big model, a target big model matching the new prompt word of the current round is screened out from the pre-set multiple big models. Specifically, the new prompt word is compared with the prompt word conditions of the remaining big models, and the big model corresponding to the prompt word condition that the new prompt word meets is used as the target big model, or the business arrangement information is queried to determine the prompt word template corresponding to the new prompt word, and the big model associated with the prompt word template is used as the target big model. Then the new prompt word is input into the target big model, and the new guidance information and new intention information output by the target big model are obtained; and it is determined whether the new intention information meets the guidance condition.

It should be noted that there are corresponding workflows in multiple business fields such as insurance, logistics, e-commerce, and finance. Business knowledge in multiple business fields is obtained, and the big model is pre-trained using the business knowledge in the same business field to obtain big models corresponding to different business fields. Each big model has the business knowledge of its own business field.

When the prompt word does not match the large model, screening out the target large model from multiple large models can improve the flexibility of large model output and meet different large model analysis needs.

According to another reference embodiment of the present invention, before obtaining the new guidance information and new intent information output by the large model of the current round, historical guidance information associated with the user is obtained. Specifically, the historical guidance information is the guidance information previously generated and returned to the user, and the historical guidance information has helped the user return to the workflow and continue to process the workflow. The execution subject of the embodiment of the present invention has a short-term memory function, and stores the guidance information generated previously. When it is necessary to generate new guidance information and new intent information, the execution subject of the embodiment of the present invention obtains the historical guidance information of the corresponding user, that is, the guidance information previously generated and returned to the user, combines the historical guidance information with the termination information, and inputs the combined information into the large model, so that the large model outputs new guidance information and new intent information according to the new prompt words and historical guidance information.

Exemplarily, the last time user F processed the workflow, he actively asked the executive body of the embodiment of the present invention "How to buy accident insurance". At this time, the workflow was terminated and the user input was only text. The executive body of the embodiment of the present invention analyzed the guidance information and intention information based on the termination information such as the user input text and the large model, and stored the guidance information as historical guidance information; when user F clicks the pop-up window to close this time, the termination information includes: the user's action of closing the page and the page that was stopped before closing. Then the executive body of the embodiment of the present invention obtains the corresponding historical guidance information of the user, that is, the user wanted to buy accident insurance last time, but the page was abnormally closed when processing the workflow of life insurance this time. It is determined that the user has the idea of buying life insurance, and the corresponding guidance information is generated to guide the user to return to the workflow of life insurance.

Generating new guidance information based on historical guidance information previously output to the user can further improve the accuracy of the guidance information and improve the efficiency of workflow processing.

According to another reference embodiment of the present invention, multiple mechanisms for triggering workflow termination are pre-set, and each mechanism for triggering workflow termination corresponds to a method for obtaining workflow termination information. Specifically, the execution subject of the embodiment of the present invention generates a page component iframe and a dialog box in the front-end page, and then displays a sub-business page (i.e., workflow page) in the iframe. The user processes the workflow in the sub-business page. When the user encounters a problem, the user asks the question in the dialog box, which triggers the workflow termination mechanism. The front-end page sends a dialogue request to the execution subject of the embodiment of the present invention through postMessage, and then the execution subject of the embodiment of the present invention determines the user's dialogue information based on the received dialogue request, and uses the dialogue information as the workflow termination information.

Alternatively, the execution subject of the embodiment of the present invention monitors the events occurring on the front-end page, and in response to detecting the occurrence of the target event, determines whether the target event belongs to a preset termination event set, and if the target event belongs to the termination event set, determines the event information of the target event, and uses the information included in the event information as the termination information of the workflow. The events in the termination event set include: page output abnormal information, page closing without successful processing, page loading failure, etc.

The user asks questions to the execution subject of the embodiment of the present invention or requests a dialogue, and detects that the workflow is abnormally exited or abnormally interrupted due to certain operations of the user, which can reduce the user's usage difficulty and improve the user's efficiency in processing the workflow.

FIG4 is a schematic diagram of a technical architecture of workflow recovery according to a reference embodiment of the present invention. Exemplarily, as shown in FIG4, the execution body 401 of the embodiment of the present invention includes: a dialog front end, a dialog business back end, a large model dialog back end, a business orchestration management and a dialog orchestration management module. Specifically, a business component iframe is generated in the dialog front end, and a sub-business front end page is displayed in the iframe. Each sub-business front end page corresponds to a link in the workflow (i.e., a workflow page). The user executes the business process in the sub-business front end page, and then asks a question in the dialog front end page. The sub-business front end synchronizes business information to the dialog front end through postMessage. The dialog front end, according to the user's request, calls up the corresponding sub-business front end, so that the user returns to the previously terminated workflow. The sub-business front end and the sub-business back end exchange business information internally, and the sub-business back end synchronizes business information with the dialog business back end, for example, the abnormal information during the execution of the synchronized sub-business, the workflow execution results, etc. The dialog front end reports the front end business status to the dialog business back end. The conversation business backend is used to determine the termination scenario, and the large model conversation backend is used to determine the guidance scenario, and generate guidance information based on the termination information, and then return the guidance information to the user through the conversation business backend and the dialog frontend. Business choreography management is used to manage the termination scenario, the business logic corresponding to the termination scenario, etc. Conversation choreography management is used to manage the guidance scenario corresponding to the termination scenario, generate the large model conversation chain, etc. Business choreography management and conversation choreography management dynamically configure the conversation business backend and the large model conversation backend.

FIG5 is a schematic diagram of the main process of workflow recovery in multi-module collaboration according to a reference embodiment of the present invention. Exemplarily, as shown in FIG5, the execution subject of the embodiment of the present invention includes: a page dialogue receiving module, a page event monitoring module, a business result receiving module, a prompt word generation module, a dialogue business processing module, a dialogue assembly module, a page dialogue display module and other modules. Among them, the page dialogue receiving module is used to receive the dialogue request sent by the user and trigger the workflow suspension mechanism process. The page event monitoring module is used to monitor the page event and trigger the workflow suspension process. The business result receiving module is used to receive the execution result of the business logic corresponding to the workflow and trigger the workflow suspension process in the event of abnormal information. The prompt word generation module is used to obtain the suspension information, convert the suspension scene into the guidance scene, determine the prompt word template, and generate the prompt word. The dialogue business processing module is used to generate the guidance information according to the prompt word, the historical dialogue between the execution subject of the embodiment of the present invention and the user, the historical guidance information, the suspension information, etc. The dialogue assembly module is used to assemble the historical dialogue; the page dialogue display module is used to display the guidance information, have a dialogue with the user, answer questions for the user, and help the user return to the workflow.

Fig. 6 is a schematic diagram of the main process of a method for workflow recovery according to a reference embodiment of the present invention. As shown in Fig. 6, the method for workflow recovery may include:

Step S601, in response to receiving a dialog request, determining the dialog information of the user according to the dialog request, and using the dialog information as the termination information of the workflow;

Step S602, determining the workflow termination scenario according to the termination information;

Step S603, determining the service logic corresponding to the suspension scenario according to the preset service arrangement information;

Step S604, executing the business logic, obtaining the execution result of the business logic, and determining the guiding scenario corresponding to the suspension scenario according to the execution result;

Step S605, updating the prompt word template corresponding to the guidance scene according to the suspension information, and using the updated prompt word template as the prompt word;

Step S606, inputting the prompt word into the preset macro model, so that the macro model generates workflow guidance information according to the prompt word;

Step S607, the guidance information is displayed on a preset front-end page, so that the user processing the workflow can return to the process node included in the workflow according to the guidance information and continue processing the workflow.

The specific implementation content of the workflow recovery method of the above-mentioned reference embodiment of the present invention has been described in detail in the above workflow recovery method, so the repeated content will not be described again here.

According to a second aspect of an embodiment of the present invention, a device for recovering a workflow is provided.

FIG. 7 is a schematic diagram of main modules of a device for workflow recovery according to an embodiment of the present invention. As shown in FIG. 7 , a device 700 for workflow recovery mainly includes:

The first generating module 701 is used to generate corresponding prompt words according to the termination information of the workflow in response to the acquisition;

The second generating module 702 is used to input the prompt word into a preset large model, so that the large model generates workflow guidance information according to the prompt word;

The guidance module 703 is used to display the guidance information on a preset front-end page, so that the user processing the workflow can return to the process node included in the workflow according to the guidance information and continue processing the workflow.

According to a reference embodiment of the present invention, a corresponding prompt word is generated according to the termination information, including: determining the termination scenario of the workflow according to the termination information; determining the guide scenario corresponding to the termination scenario according to pre-set business orchestration information; updating the prompt word template corresponding to the guide scenario according to the termination information, and using the updated prompt word template as the prompt word.

According to another reference embodiment of the present invention, a guiding scenario corresponding to the termination scenario is determined according to pre-set business orchestration information, including: determining the business logic corresponding to the termination scenario according to the business orchestration information; executing the business logic to obtain the execution result of the business logic; and determining the guiding scenario corresponding to the termination scenario according to the execution result.

According to another reference embodiment of the present invention, the guidance module 703 is further used to: before displaying the guidance information on a preset front-end page, determine the intention information associated with the guidance information, and determine whether the intention information meets the preset guidance condition; if the intention information meets the preset guidance condition, loop the following steps to determine the guidance information to be displayed on the front-end page:

Use the intent information as the intent information of the first round, determine the user purpose information corresponding to the intent information of the current round, and generate new prompt words for the current round according to the user purpose information; determine the big model of the current round according to the big model, input the new prompt words of the current round into the big model of the current round, and obtain the new intent information and new guidance information output by the big model of the current round; when the new intent information of the current round does not meet the guidance conditions, the loop is terminated, and the new guidance information of the current round is used as the guidance information displayed on the front-end page; when the new intent information of the current round meets the guidance conditions, the new intent information of the current round is used as the intent information for the next round.

According to another reference embodiment of the present invention, the guiding module 703 is also used for: in response to determining that the new prompt word of the current round matches the big model, using the big model as the big model of the current round; in response to determining that the new prompt word does not match the big model, screening out a target big model that matches the new prompt word of the current round from a plurality of pre-set big models; using the target big model as the big model of the current round.

According to another reference embodiment of the present invention, the guidance module 703 is also used to: obtain historical guidance information associated with the user; input the historical guidance information into the big model of the current round, so that the big model of the current round outputs new guidance information and new intention information of the current round according to the new prompt words of the current round and the historical guidance information.

According to a reference embodiment of the present invention, the termination information of the workflow is obtained, including: in response to receiving a dialogue request, determining the user's dialogue information according to the dialogue request, and using the dialogue information as the termination information of the workflow; or, in response to detecting the occurrence of a target event, determining whether the target event belongs to a preset termination event set, and in the case that the target event belongs to the termination event set, determining the event information of the target event, and using the event information as the termination information of the workflow.

It should be noted that the specific implementation content of the device for workflow recovery in the embodiment of the present invention has been described in detail in the above workflow recovery method, so the repeated content will not be described again here.

According to the technical solution of the embodiment of the present invention, the termination information of the workflow is parsed based on the big model to obtain the corresponding guidance information, so that the user can return to the process node of the workflow according to the guidance information and continue to process the workflow, which can improve the processing efficiency of the workflow, reduce the difficulty of users processing the workflow, and improve the user experience; according to the business arrangement information, the guidance scenario corresponding to the termination scenario is determined, and the prompt word is generated according to the guidance scenario, which can flexibly and accurately generate the prompt word; the business logic corresponding to the termination scenario is executed, and the guidance scenario is determined according to the execution result of the business logic, which can flexibly and efficiently determine the guidance scenario; according to the intention information associated with the guidance information, it is determined whether it is necessary to generate new guidance information and intention information again, and after multiple big model outputs, the final return is obtained. The guidance information given to the user can accurately determine the user's intention and purpose, facilitate the provision of accurate guidance information to the user, and help the user quickly return to the process node of the workflow; when the prompt word does not match the large model, the target large model is screened out from multiple large models, which can improve the flexibility of the large model output and meet different large model analysis needs; based on the historical guidance information previously output to the user, new guidance information is generated, which can further improve the accuracy of the guidance information and improve the workflow processing efficiency; the triggering timing of the method includes: the user asks questions to the execution subject of the embodiment of the present invention or requests a dialogue, and detects that the workflow is abnormally exited or abnormally interrupted due to certain operations of the user, which can reduce the user's usage difficulty and improve the user's efficiency in processing the workflow.

According to a third aspect of an embodiment of the present invention, there is provided an electronic device, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method provided by the first aspect of the embodiment of the present invention.

According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method provided by the first aspect of the embodiment of the present invention is implemented.

FIG. 8 shows an exemplary system architecture 800 to which the method for workflow recovery or the apparatus for workflow recovery according to an embodiment of the present invention can be applied.

As shown in Fig. 8, system architecture 800 may include terminal devices 801, 802, 803, a network 804 and a server 805. Network 804 is used to provide a medium for communication links between terminal devices 801, 802, 803 and server 805. Network 804 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

The user can use the terminal devices 801, 802, 803 to interact with the server 805 through the network 804 to receive or send messages, etc. Various communication client applications can be installed on the terminal devices 801, 802, 803, such as workflow recovery applications, workflow query applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

The terminal devices 801 , 802 , and 803 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

Server 805 can be a server that provides various services, such as a background management server that provides support for workflow recovery requests sent by upstream terminal devices 801, 802, and 803 (for example only). The background management server can generate corresponding prompt words according to the termination information obtained from the workflow in response to the termination information; input the prompt words into a pre-set large model so that the large model generates workflow guidance information according to the prompt words; display the guidance information to a pre-set front-end page so that the user processing the workflow can return to the process node included in the workflow according to the guidance information and continue processing the workflow; and feedback the workflow recovery situation (for example only) to the terminal device.

It should be noted that the workflow recovery method provided in the embodiment of the present invention is generally executed by the server 805, and accordingly, the workflow recovery device is generally set in the server 805. The workflow recovery method provided in the embodiment of the present invention can also be executed by the terminal devices 801, 802, and 803, and accordingly, the workflow recovery device can be set in the terminal devices 801, 802, and 803.

It should be understood that the number of terminal devices, networks and servers in Figure 8 is only for illustration purposes and any number of terminal devices, networks and servers may be provided as required.

Referring to Figure 9, a schematic diagram of a computer system 900 suitable for implementing a terminal device of an embodiment of the present invention is shown. The terminal device shown in Figure 9 is only an example and should not limit the functions and scope of use of the embodiment of the present invention.

As shown in FIG9 , the computer system 900 includes a central processing unit (CPU) 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage portion 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the system 900 are also stored. The CPU 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input/output (I/O) interface 905 is also connected to the bus 904.

The following components are connected to the I/O interface 905: an input section 906 including a keyboard, a mouse, etc.; an output section 907 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, a modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the I/O interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 910 as needed, so that a computer program read therefrom is installed into the storage section 908 as needed.

In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 909, and/or installed from the removable medium 911. When the computer program is executed by the central processing unit (CPU) 901, the above-mentioned functions defined in the system of the embodiment of the present invention are executed.

It should be noted that the computer-readable medium shown in the embodiment of the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In an embodiment of the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In an embodiment of the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

The modules involved in the embodiments of the present invention may be implemented by software or hardware. The modules described may also be set in a processor, for example, it may be described as: a processor includes a first generation module, a second generation module, and a boot module, wherein the names of these modules do not constitute a limitation on the modules themselves under certain circumstances, for example, the first generation module may also be described as a "module for generating prompt words according to the termination information".

As another aspect, an embodiment of the present invention further provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently and not be assembled into the device. The above computer-readable medium carries one or more programs. When the above one or more programs are executed by a device, the device implements the following method: in response to obtaining the termination information of the workflow, generating a corresponding prompt word according to the termination information; inputting the prompt word into a pre-set large model so that the large model generates the guidance information of the workflow according to the prompt word; displaying the guidance information to a pre-set front-end page so that the user who processes the workflow can return to the process node included in the workflow according to the guidance information and continue to process the workflow.

According to the technical solution of the embodiment of the present invention, the termination information of the workflow is parsed based on the big model to obtain the corresponding guidance information, so that the user can return to the process node of the workflow according to the guidance information and continue to process the workflow, which can improve the processing efficiency of the workflow, reduce the difficulty of users processing the workflow, and improve the user experience; according to the business arrangement information, the guidance scenario corresponding to the termination scenario is determined, and the prompt word is generated according to the guidance scenario, which can flexibly and accurately generate the prompt word; the business logic corresponding to the termination scenario is executed, and the guidance scenario is determined according to the execution result of the business logic, which can flexibly and efficiently determine the guidance scenario; according to the intention information associated with the guidance information, it is determined whether it is necessary to generate new guidance information and intention information again, and after multiple big model outputs, the final return is obtained. The guidance information given to the user can accurately determine the user's intention and purpose, facilitate the provision of accurate guidance information to the user, and help the user quickly return to the process node of the workflow; when the prompt word does not match the large model, the target large model is screened out from multiple large models, which can improve the flexibility of the large model output and meet different large model analysis needs; based on the historical guidance information previously output to the user, new guidance information is generated, which can further improve the accuracy of the guidance information and improve the workflow processing efficiency; the triggering timing of the method includes: the user asks questions to the execution subject of the embodiment of the present invention or requests a dialogue, and detects that the workflow is abnormally exited or abnormally interrupted due to certain operations of the user, which can reduce the user's usage difficulty and improve the user's efficiency in processing the workflow.

The above specific implementations do not constitute a limitation on the protection scope of the embodiments of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the embodiments of the present invention should be included in the protection scope of the embodiments of the present invention.

Claims (11)

1. A method of workflow restoration, comprising:

Responding to the obtained suspension information of the workflow, and generating a corresponding prompt word according to the suspension information;

inputting the prompt word into a preset large model, and enabling the large model to generate the guide information of the workflow according to the prompt word;

and displaying the guide information to a preset front-end page, so that a user processing the workflow returns to a flow node included in the workflow according to the guide information, and continues to process the workflow.

2. The method of claim 1, wherein generating the corresponding alert word based on the suspension information comprises:

Determining a suspension scene of the workflow according to the suspension information;

determining a guide scene corresponding to the suspension scene according to preset service arrangement information;

and updating the prompt word template corresponding to the guide scene according to the suspension information, and taking the updated prompt word template as the prompt word.

3. The method according to claim 2, wherein determining the guidance scenario corresponding to the suspension scenario according to the preset service arrangement information includes:

Determining service logic corresponding to the suspension scene according to the service arrangement information;

Executing the service logic and obtaining an execution result of the service logic;

and determining a guide scene corresponding to the suspension scene according to the execution result.

4. The method of claim 1, wherein prior to presenting the guidance information to the pre-set front-end page, the method further comprises:

determining intention information associated with the guide information, and judging that the intention information does not accord with preset guide conditions;

And under the condition that the intention information accords with a preset guiding condition, circularly executing the following steps to determine guiding information displayed to the front-end page:

The intention information is used as intention information of a first round, user destination information corresponding to the intention information of the current round is determined, and a new prompt word of the current round is generated according to the user destination information; determining a large model of the current round according to the large model, inputting a new prompt word of the current round into the large model of the current round, and acquiring new intention information and new guide information output by the large model of the current round; when the new intention information of the current round does not accord with the guide condition, the circulation is terminated, and the new guide information of the current round is used as the guide information displayed to the front page; and taking the new intention information of the current round as the intention information of the next round under the condition that the new intention information of the current round meets the guide condition.

5. The method of claim 4, wherein determining a large model of a current round from the large model comprises:

responding to the judgment that the new prompt word of the current round is matched with the large model, and taking the large model as the large model of the current round;

responding to the judgment that the new prompting words of the current round are not matched with the large models, and screening target large models matched with the new prompting words of the current round from a plurality of preset large models; and taking the target large model as a large model of the current round.

6. The method of claim 5, wherein prior to obtaining new guidance information and new intent information for the large model output for the current round, the method further comprises:

acquiring history guidance information associated with the user;

And inputting the history guide information into a large model of the current round, so that the large model of the current round outputs new guide information and new intention information of the current round according to the new prompt word of the current round and the history guide information.

7. The method of claim 1, wherein obtaining the workflow suspension information comprises:

In response to receiving a dialogue request, determining dialogue information of the user according to the dialogue request, and taking the dialogue information as the stopping information of the workflow;

Or in response to detecting that a target event occurs, judging whether the target event belongs to a preset suspension event set, and determining event information of the target event under the condition that the target event belongs to the suspension event set, wherein the event information is used as suspension information of the workflow.

8. An apparatus for workflow restoration, comprising:

The first generation module is used for responding to the obtained suspension information of the workflow and generating corresponding prompt words according to the suspension information;

The second generation module is used for inputting the prompt word into a preset large model, so that the large model generates the guide information of the workflow according to the prompt word;

and the guiding module is used for displaying the guiding information to a preset front-end page, so that a user processing the workflow returns to a flow node included in the workflow according to the guiding information and continues to process the workflow.

9. An electronic device, comprising:

one or more processors;

storage means for storing one or more programs,

The one or more processors implement the method of any of claims 1-7 when the one or more programs are executed by the one or more processors.

10. A computer readable medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method according to any of claims 1-7.

11. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any of claims 1-7.