Interaction method, apparatus, system, electronic device, and storage medium

Abstract

The application provides an interaction method, device, system, electronic equipment and storage medium, and relates to the technical field of computers, and the method comprises the following steps: obtaining a dialogue request; obtaining target user information corresponding to a user request end and a business type of a target task according to the dialogue request; determining the plug-in state of a user audit module according to the target user information and the business type; obtaining a plurality of inference models corresponding to the target task according to the plug-in state and the business type, processing the target task based on the plurality of inference models, and obtaining a plurality of processing results; auditing and configuring the plurality of processing results according to the plug-in state, obtaining response information corresponding to the dialogue request, and pushing the response information to the user request end. The application improves the completeness, accuracy and flexibility of interaction.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to an interactive method, apparatus, system, electronic device, and storage medium. Background Technology

[0002] With the development of the internet and big data, and the emergence of ChatGPT (Chat-based Generative Pre-trained Transformer), the AI ​​(Artificial Intelligence) industry is experiencing explosive growth. Consequently, a large number of AI-related large models have been developed to meet various business needs, and for each of these models, a series of generative artificial intelligence systems have been derived to interact with users and help them generate corresponding solutions to a range of problems.

[0003] Generative AI systems typically collect, clean, and process data from various systems, then train the data to generate a single, large model. Based on user input, the system retrieves results from this single model. However, the results of this single model are often heavily influenced by the training data and the model's structure, which can compromise the accuracy of the interactive responses returned to the user.

[0004] Therefore, there is an urgent need for an interaction method, device, electronic device, and storage medium to solve the above problems. Summary of the Invention

[0005] This invention provides an interaction method, device, system, electronic device, and storage medium to address the shortcomings of existing artificial intelligence systems where the results of large models are often greatly affected by training data and model structure, leading to the inability to guarantee the correctness of interaction, and to achieve accurate interaction.

[0006] This invention provides an interaction method, comprising:

[0007] Obtain a dialogue request; the dialogue request includes a processing request for the target task input by the user request terminal;

[0008] Based on the dialogue request, obtain the target user information corresponding to the user requesting end and the business type of the target task;

[0009] Based on the target user information and the business type, determine the plug-in / plug-out status of the user review module;

[0010] Based on the plug-in / plug-out status and the service type, multiple inference models corresponding to the target task are obtained, and the target task is processed based on the multiple inference models to obtain multiple processing results;

[0011] Based on the plug-in / plug-out status, multiple processing results are reviewed and configured to obtain response information corresponding to the dialogue request, and the response information is pushed to the user requesting end.

[0012] According to an interaction method provided by the present invention, the step of obtaining target user information corresponding to the user requesting end and the business type of the target task based on the dialogue request includes:

[0013] The dialog request is format-converted to obtain a converted dialog request.

[0014] Based on the converted dialogue request, obtain the target user information and the intent information of the dialogue request;

[0015] Based on the intent information, the business type is obtained by matching it in the business information table of the database.

[0016] According to an interaction method provided by the present invention, determining the plug-in / plug-out status of the user review module based on the target user information and the service type includes:

[0017] In the database's audit configuration data table, find the user audit configuration information associated with the target user information;

[0018] If user review configuration information associated with the target user information is found, then the user review configuration information is searched for to find first user review authorization information associated with the target user information and second user review authorization information associated with the business type.

[0019] If it is determined that a first user review authorization information associated with the target user information and a second user review authorization information associated with the business type are found, the plug-in / plug-out status of the user review module is updated to the enabled status.

[0020] According to an interaction method provided by the present invention, the method further includes:

[0021] Obtain a first configuration request; the first configuration request includes a configuration request for the audit configuration data table input through the user configuration terminal;

[0022] Based on the first configuration request, obtain the object to be configured and the user approval configuration information associated with the object to be configured; the object to be configured includes the user to be configured and / or the service to be configured.

[0023] Locate the object to be configured in the audit configuration data table;

[0024] If the object to be configured is not found, the object to be configured and the user review configuration information associated with the object to be configured are added to the review configuration data table in the form of a mapping relationship.

[0025] According to an interaction method provided by the present invention, obtaining multiple inference models corresponding to the target task based on the plug-in / plug-out state and the service type includes:

[0026] If the plug-in / plug-out status is determined to be enabled, the target audit type is determined in the audit configuration data table of the database according to the service type.

[0027] Based on the target audit type, multiple inference models are obtained from the business configuration data table in the database.

[0028] According to an interaction method provided by the present invention, the step of reviewing and configuring multiple processing results based on the plugging / unplugging state to obtain response information corresponding to the dialogue request, and pushing the response information to the user requesting end, includes:

[0029] When the plugging / unplugging state is enabled, a target audit task is generated based on multiple processing results;

[0030] Store the target review task in the review queue;

[0031] The target review task is read from the review queue according to the trigger time and the storage order of the target review task.

[0032] Upon receiving the target review task, information to be reviewed is generated based on the target review task and the dialogue request;

[0033] The information to be reviewed is pushed to the user configuration terminal via a server push event connection with the user configuration terminal;

[0034] Obtain a second configuration request; the second configuration request includes a request entered when reviewing and configuring multiple processing results based on the information to be reviewed displayed on the user configuration terminal;

[0035] Based on the second configuration request, obtain the target configuration operation;

[0036] Based on the target configuration operation, configure multiple processing results to obtain response information corresponding to the dialogue request;

[0037] The response information corresponding to the dialogue request is pushed to the user requesting client through a server push event connection with the user requesting client.

[0038] According to an interaction method provided by the present invention, the step of reviewing and configuring multiple processing results based on the plugging / unplugging state to obtain response information corresponding to the dialogue request includes...

[0039] When the user review module is in a disabled state, multiple processing results are reviewed and configured according to a preset review algorithm to obtain the response information corresponding to the dialogue request.

[0040] The present invention also provides an interactive device, including a front end and a server end;

[0041] The server includes a processing module, an audit verification module, an audit type determination module, an inference module, and an audit push module;

[0042] The front end is used to obtain dialogue requests; the dialogue requests include processing requests for target tasks input by the user request terminal;

[0043] The processing module is used to obtain the target user information corresponding to the user requesting end and the business type of the target task according to the dialogue request.

[0044] The audit verification module is used to determine the plug-in / plug-out status of the user audit module based on the target user information and the business type.

[0045] The audit type determination module is used to obtain multiple inference models corresponding to the target task based on the plug-in / plug-out status and the business type.

[0046] The reasoning module is used to process the target task based on multiple reasoning models to obtain multiple processing results;

[0047] The review and push module is used to review and configure multiple processing results according to the plug-in / plug-out status, obtain response information corresponding to the dialogue request, and push the response information to the user request terminal.

[0048] The present invention also provides an interactive system, including the interactive device as described above, at least one user request terminal, at least one user configuration terminal, and a database;

[0049] The interactive device establishes communication with each of the user request terminals and each of the user configuration terminals through server push event connections.

[0050] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the interaction method as described above.

[0051] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the interaction method as described above.

[0052] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the interaction method as described above.

[0053] The interactive method, apparatus, system, electronic device, and storage medium provided by this invention obtain target user information and target task business type corresponding to the user request terminal through dialogue requests. Based on the target user information and business type, the plug-in / plug-out status of the user review module is determined. This allows for the adaptive acquisition of multiple inference models and the review and configuration of multiple processing results based on the plug-in / plug-out status of the user review module. By adaptively performing human-computer interactive review configuration or machine-autonomous review configuration based on the inference results of multiple inference models, the final response result is obtained. This not only improves the completeness of the interaction but also enhances its accuracy and flexibility. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0055] Figure 1 This is a flowchart illustrating the interaction method provided by the present invention;

[0056] Figure 2 This is a schematic diagram of the structure of the interactive system provided by the present invention;

[0057] Figure 3 This is a schematic diagram of the interaction process between the interactive device and the user configuration terminal provided by the present invention.

[0058] Figure 4 This is a schematic diagram of the user review configuration information retrieval process provided by the present invention;

[0059] Figure 5 This is one of the flowcharts illustrating the configuration of the audit configuration data table provided by the present invention;

[0060] Figure 6 This is the second flowchart illustrating the configuration of the audit configuration data table provided by this invention;

[0061] Figure 7This is a schematic diagram of the process for obtaining the target review type provided by the present invention;

[0062] Figure 8 This is one of the flowcharts illustrating SSE interaction provided by the present invention;

[0063] Figure 9 This is the second flowchart of SSE interaction provided by the present invention;

[0064] Figure 10 This is a schematic diagram of the structure of the interactive device provided by the present invention;

[0065] Figure 11 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0067] The following is combined with Figures 1-9 The interactive method of the present invention is described.

[0068] Figure 1 This is a flowchart illustrating the interaction method provided in this embodiment. This method can be applied to interactive systems in other systems requiring interaction, such as artificial intelligence systems and office software. The artificial intelligence system refers to a technology that generates text, images, audio, video, instructions, etc., based on algorithms, models, and rules, providing generative artificial intelligence products or services. The following description uses an artificial intelligence system as an example to illustrate the interaction method provided in this embodiment. Equivalent substitutions can be made for other systems, and will not be elaborated upon here.

[0069] The interactive system can be based on the Spring framework, an open-source application framework that supports Java application development by providing infrastructure support. The execution entity of this method is the interactive device within the interactive system. This interactive device can be a chat display system or other interactive platforms. The chat display system is for company reception and presentation purposes only and is not open to the public.

[0070] Figure 2 This is a schematic diagram of the interactive system structure; such as Figure 2As shown, in order to achieve adaptive switching between human-computer interactive review and machine autonomous review, the system embeds various user review (or manual review, or human-computer interactive review) configuration modules and user review function modules, thereby achieving different processes in different scenarios; specifically, the interactive system includes at least an interactive device, at least one user request terminal, at least one user configuration terminal, and a database;

[0071] The interaction between the interactive device and the user request terminal, as well as between the interactive device and the user configuration terminal, can be conducted using SSE (Server Send Event). Compared with the polling mechanism (that is, the mechanism of requesting and obtaining results at a fixed frequency), this improves the real-time performance of the interaction and enhances the stability of the interactive device.

[0072] The database referred to may be a MySQL database used to store interactive data and configuration information. It may be configured with corresponding databases on the interactive device, the user request end, and the user configuration end, or it may be configured with a single database on the interactive device, the user request end, and the user configuration end. This embodiment does not specifically limit this.

[0073] The interactive device is used for processing and displaying interactive information; the user configuration terminal is used for configuring the content to be reviewed and for reviewing and configuring the reasoning results.

[0074] For interactive devices (such as display systems), this includes the front-end, server-side, and database;

[0075] The interactive functionality of the front-end of the interactive device can be implemented through a WEB (World Wide Web) page for user interaction, specifically including the display of input and output information. The server-side of the interactive device includes an input / output module, a configuration module, a task timing module, a processing module, an audit verification module, an audit type determination module, a pluggable audit module, an audit queue module, and an audit push module. The input / output module processes user input and formats the output content. The configuration module receives configuration requests from the user's configuration end and configures various information in the database, such as the intent information and business type configuration in the business information table, the user information, business type, and user audit configuration information configuration in the audit configuration data table, the business type and audit type configuration in the audit configuration data table, and the audit type and inference model configuration in the business configuration data table. The task timing module automatically processes tasks at regular intervals according to rules. The processing module performs various data processing operations, such as retrieving target user information and the business type of the target task based on user requests. The pluggable audit module stores the relevant judgment and processing logic for pluggable audits; the audit type determination module is used for pluggable configuration of the audit type for different business types; the audit verification module calls the logic in the pluggable audit module to make decisions on whether to conduct human-computer interactive audits or machine-autonomous audits, that is, based on user information and the business type of the task, it determines whether human-computer interactive audits are currently enabled. If not enabled, it directly calls the inference based on the current logic and obtains the result. If enabled, it needs to interact with the audit type determination module to obtain the inference model to be called and obtain the result; the audit queue module stores user audit tasks; and the audit push module pushes the response information after the audit configuration to the front end.

[0076] For user configuration, it can also include the front-end, server, and database.

[0077] The interactive functions of the user configuration front-end can also be implemented through a web page, which is used by administrators to configure approval information and by administrators to conduct manual approval.

[0078] The user configuration server includes a pluggable configuration module, a content moderation module, a moderation queue module, and a moderation push module. The pluggable configuration module allows administrators to configure pluggable information, such as user information, business type and user moderation configuration information in the moderation configuration data table, business type and moderation type configuration in the moderation configuration data table, and moderation type and inference model configuration in the business configuration data table. This allows administrators to easily and flexibly configure various moderation-related items in real-time, storing the configurations in a database for interactive devices to access and retrieve in various business logics, thus achieving pluggable responses to different scenarios. The content moderation module allows administrators to retrieve information to be reviewed from the review list, view multiple results returned by the inference model, modify the results, and submit the final result, enabling moderation and configuration of the processing results. The moderation queue module handles user moderation tasks. The moderation push module pushes the configured response information to the front end.

[0079] The interactive device interacts with both the user configuration terminal and the user request terminal. This allows the interactive device to update the database with pluggable configuration information for the required review information, as configured by the administrator through the user configuration terminal. After a user submits a dialogue request on the user request terminal, the interactive device can perform inference processing. Based on the configuration for whether to perform human-computer interaction review, it determines whether the inference result should be placed in the user review queue. If human-computer interaction review is required, the content to be reviewed is retrieved from the review queue module via SSE (Server Send Event) and pushed to the front-end page of the user configuration terminal. The administrator then processes and submits the request. Upon receiving the final response, the interactive device pushes the result to the front-end interface of the user request terminal via SSE or displays it on the front-end page of the interactive device. This pluggable human-computer interaction review method configures the review process for users and various business applications, allowing for interactive content review on the user configuration terminal or self-review on the user's own end. Finally, a more accurate and reasonable result is returned to the user request terminal for the user to view, thereby improving the accuracy and flexibility of the interaction.

[0080] like Figure 1 As shown, the method includes the following steps:

[0081] Step 101, obtain a dialogue request; the dialogue request includes a processing request for the target task input by the user request terminal;

[0082] The target task is the task that needs to be processed, such as a question-and-answer task, a drawing task, or a poetry writing task. This embodiment does not specifically limit this task.

[0083] Optionally, when a user needs to process a target task, they can input a dialogue request for task processing on the front-end interface of the user requesting end; the dialogue request is information input through command line interface, graphical interface, touch input, drop-down selection input, voice input, gesture input, visual input, brain-computer input, etc., and this embodiment does not specifically limit it.

[0084] Step 102: Based on the dialogue request, obtain the target user information corresponding to the user requesting end and the business type of the target task;

[0085] Optionally, after receiving a dialogue request, the interactive device may invoke the processing unit to perform the following operations:

[0086] The dialogue request can be parsed directly to obtain the target user information and the business type of the target task; or the dialogue request can be pre-processed, such as format conversion and intent analysis, so as to parse the processed dialogue request to obtain the target user information and the business type of the target task. This embodiment does not specifically limit this.

[0087] Step 103: Determine the plug-in / plug-out status of the user review module based on the target user information and the business type;

[0088] It should be noted that, in order to improve the versatility and flexibility of the interactive device, the user review module is designed to be pluggable. That is, when the user review module is needed, it is activated to interactively review the inference results generated by the generative artificial intelligence system; when the user review module is not needed, it is deactivated, and the interactive device reviews the inference results autonomously, thus realizing the pluggable management of the user review module.

[0089] The term "plug-and-play" refers to the ability to enable or disable a device or function through configuration or operation. In this embodiment, the plug-and-play feature primarily involves enabling or disabling the user review module through configuration, as well as making plug-and-play changes to the content related to the review type.

[0090] In addition, different user information and business types correspond to different plug-in / plug-out states.

[0091] Optionally, after obtaining the target user information and business type, the audit verification module can be invoked to perform the following steps:

[0092] By combining target user information and business type, we can obtain the plug-in / plug-out status of the user review module.

[0093] The methods for determining the plug-in / plug-out status of the user review module include: obtaining the plug-in / plug-out status of the user review module based on pre-configured plug-in / plug-out review logic; or identifying the plug-in / plug-out status of the user review module based on a pre-trained recognition model. For example, obtaining the corresponding user review configuration information based on the target user information, and then determining whether user review configuration information is configured for the user and whether user review configuration information is configured for the target business type, thereby determining the plug-in / plug-out status of the user review module; or collecting a large amount of sample user information and sample business types, as well as corresponding plug-in / plug-out status labels, and training a neural network model to obtain a recognition model that can accurately identify the plug-in / plug-out status of the user review module based on user information and business type; subsequently, inputting the target user information and business type into the recognition model to obtain the plug-in / plug-out status of the user review module.

[0094] Step 104: Based on the plug-in / plug-out status and the service type, obtain multiple inference models corresponding to the target task, process the target task based on the multiple inference models, and obtain multiple processing results;

[0095] Different plug-in / plug-out states correspond to different inference model acquisition methods. For example, when the plug-in / plug-out state is in the enabled state, multiple inference models corresponding to the target task can be obtained based on the audit configuration information in the audit configuration data table in the database, such as the audit type. When the plug-in / plug-out state is in the disabled state, multiple inference models corresponding to the target task can be obtained directly based on the business type.

[0096] Optionally, after obtaining the plug-in / plug-out status and business type, the following steps can be performed by calling the audit type determination module:

[0097] Based on the plug-in / plug-out status, the corresponding inference model acquisition method is obtained. Then, based on the corresponding inference model acquisition method, multiple inference models corresponding to the target task are searched and obtained according to the business type.

[0098] Multiple inference models are invoked to process the target task synchronously, resulting in multiple processing outcomes.

[0099] Step 105: Based on the plugging / unplugging status, review and configure multiple processing results to obtain response information corresponding to the dialogue request, and push the response information to the user requesting end.

[0100] Different plug-in / plug-out states correspond to different audit strategies. For example, when the user audit module is plugged in and out in the enabled state, the audit strategy required for the interaction process is a human-computer interactive audit strategy; when the user audit module is plugged in and out in the disabled state, the audit strategy required for the interaction process is an interactive device autonomous audit strategy. This allows the audit strategy to be adaptively determined according to the plug-in / plug-out state of the user audit module during the interaction process. That is, machine autonomous audit or human-computer interactive audit can be performed, and the two can be switched at any time, which is flexible, real-time, convenient, and efficient. This allows for different logical processing of the processing results generated during the interaction process, thereby eliminating poor-quality processing results and improving the accuracy, controllability, and adaptability of the interaction results.

[0101] Optionally, after obtaining multiple processing results, the review push module can be invoked to perform the following steps:

[0102] Based on the plug-in / plug-out status, the corresponding audit policy is obtained. Then, based on the corresponding audit policy, the multiple processing results obtained in step 104 are audited and configured. This results in pushing more accurate and reasonable response information to the user requesting end, so that the user requesting end can obtain more reasonable and accurate response information and improve the user experience.

[0103] The interaction method provided in this embodiment obtains the target user information and the business type of the target task corresponding to the user requesting end through a dialogue request. Based on the target user information and business type, the plug-in / plug-out status of the user review module is determined. This allows for the adaptive acquisition of multiple inference models and the review and configuration of multiple processing results based on the plug-in / plug-out status of the user review module. By adaptively performing human-computer interactive review configuration or machine autonomous review configuration based on the inference results of multiple inference models, the final response result is obtained. This not only improves the completeness of the interaction but also enhances the accuracy and flexibility of the interaction.

[0104] In some embodiments, step 102, which involves obtaining the target user information corresponding to the user requesting end and the business type of the target task based on the dialogue request, includes:

[0105] The dialog request is format-converted to obtain a converted dialog request.

[0106] The converted dialogue request is parsed to obtain the target user information and the intent information of the dialogue request;

[0107] Based on the intent information, the business type is obtained by matching it in the business information table of the database.

[0108] Figure 3This is a schematic diagram illustrating the interaction process between the interactive device and the user configuration terminal provided in this embodiment; as follows: Figure 3 As shown, after obtaining the dialogue request through step 101, in order to obtain the target user information and the business type of the target task more quickly and accurately, the interactive device may call the processing module to perform the following processing on the dialogue request:

[0109] The format of the dialogue request is converted to a format compatible with the processing format of the interactive device.

[0110] Then, the converted dialogue request is analyzed to obtain target user information and the intent information of the dialogue request.

[0111] The methods for obtaining target user information and intent information can be based on identifying and obtaining dialogue requests using a pre-trained language model, or by parsing the dialogue requests. For example, parsing the converted dialogue requests can extract target user information and intent information from them. Alternatively, sample dialogues, along with corresponding user tags and intent tags, can be collected to train a language model, resulting in a language model that can identify the corresponding user information and intent information based on the dialogue requests. The converted dialogue requests can then be input into the language model to output the corresponding target user information and intent information.

[0112] Next, after obtaining the intent information, the corresponding business type can be obtained by matching it in the business information table of the database.

[0113] The database's business information table is pre-configured with multiple sets of intent information and business types based on mapping relationships, and this table can also be adapted to meet actual needs. Configuration includes adding, modifying, and deleting data.

[0114] The method provided in this embodiment converts the format of the dialogue request and obtains the target user information and the intent information of the dialogue request based on the converted dialogue request. The service type is then obtained by matching the intent information, thereby better understanding the user intent, improving the efficiency of interactive processing and response speed, and thus providing users with more accurate and personalized services.

[0115] In some embodiments, determining the plug-in / plug-out status of the user review module based on the target user information and the service type in step 103 includes:

[0116] In the database's audit configuration data table, find the user audit configuration information associated with the target user information;

[0117] If user review configuration information associated with the target user information is found, then the user review configuration information is searched for to find first user review authorization information associated with the target user information and second user review authorization information associated with the business type.

[0118] If it is determined that a first user review authorization information associated with the target user information and a second user review authorization information associated with the business type are found, the plug-in / plug-out status of the user review module is updated to the enabled status.

[0119] like Figure 3 As shown, the interactive device can call the audit verification module to perform audit verification logic processing in order to determine the plug-in / plug-out status of the user audit module.

[0120] Figure 4 This is a schematic diagram of the user review configuration information retrieval process provided in this embodiment; such as Figure 3 and Figure 4 As shown, the steps for determining the plug-in / plug-out status of the user review module in step 103 include:

[0121] First, obtain information about the target users and the business type of the target business;

[0122] Next, using the target user information as the dimension, the user review configuration information associated with the target user information is retrieved from the review configuration data table in the database. Specifically, this can be done by calling the review verification module to retrieve the user review configuration information using a command such as "needVerify(userId,applicationId)". The command returns a true / false result, and the result is used to determine whether the user review configuration information associated with the target user information has been retrieved.

[0123] It should be noted that the database's audit configuration table is pre-configured with multiple sets of user information and user audit configuration information based on mapping relationships. Furthermore, this database's audit configuration table can be adaptively added to, modified, and deleted according to actual needs. The system then calls the audit verification module to retrieve the audit configuration using the retrieval configuration method, obtaining a pluggable configuration result. It should also be noted that the mapping relationship between user information and user audit configuration information can be configured based on the user's importance level and the business type's importance level. For example, for important users and important business types, user audit authorization information is configured to obtain the corresponding user audit configuration information, effectively enabling the user audit module to prevent uncontrollable and abnormal interaction information; for non-important users or non-important business types, user audit authorization information is not configured to obtain the corresponding user audit configuration information, effectively disabling the user audit module.

[0124] If no user review configuration information associated with the target user information is obtained, the user review module is determined to be disabled, meaning that the review is conducted autonomously using an interactive device, and no human-computer interactive review is performed. If user review configuration information associated with the target user information is found, meaning that the review configuration data table contains user review configuration information associated with the target user information, then it is further determined whether the target user information and the business type of the target business are associated with the user review configuration information. If they are associated with the user review configuration information, the user review module is determined to be enabled, meaning that the user review model needs to be activated for human-computer interactive review. If they are not associated with the user review configuration information, the user review module is determined to be disabled, meaning that the review is conducted autonomously using an interactive device, and no human-computer interactive review is performed.

[0125] At this point, the steps for determining whether the target user information and the business type of the target service are associated with the user review configuration information include: in the user review configuration information, searching for the first user review authorization information associated with the target user information and the second user review authorization information associated with the business type; if the first user review authorization information and the second user review authorization information are found, it indicates that the target user information and the business type of the target service are associated with the user review configuration information; if the first user review authorization information and / or the second user review authorization information are not found, it indicates that the target user information and / or the business type of the target service are not associated with the user review configuration information.

[0126] The method provided in this embodiment determines whether to enable human-computer interactive review operation for dialogue requests based on the user's review configuration information and permission settings, and updates the plug-in / plug-out status of the user's review module in real time to ensure the security and compliance of the review operation.

[0127] In some embodiments, the method further includes:

[0128] Obtain a first configuration request; the first configuration request includes a configuration request for the audit configuration data table input through the user configuration terminal;

[0129] Based on the first configuration request, obtain the object to be configured and the user approval configuration information associated with the object to be configured; the object to be configured includes the user to be configured and / or the service to be configured.

[0130] Locate the object to be configured in the audit configuration data table;

[0131] If the object to be configured is not found, the object to be configured and the user review configuration information associated with the object to be configured are added to the review configuration data table in the form of a mapping relationship.

[0132] Figure 5 This is one of the flowcharts illustrating the configuration of the audit configuration data table provided in this embodiment. Figure 6 This is the second flowchart illustrating the configuration process of the audit configuration data table provided in this embodiment; for example... Figure 5 and Figure 6 As shown, the steps for reviewing the configuration data table settings include:

[0133] Receive the configuration request, also known as the first configuration request, which is the configuration data table of the pluggable configuration module that the user inputs into the front end of the user configuration terminal.

[0134] Next, the configuration request is parsed to obtain the object to be configured and the user approval configuration information associated with the object to be configured; the object to be configured includes the user to be configured and / or the service to be configured.

[0135] If the object to be configured includes a user to be configured, the user to be configured is searched in the audit configuration data table of the database using the user to be configured as the search dimension. If the user to be configured is found, the original user audit configuration information of the user to be configured in the audit configuration data table is modified according to the user audit configuration information associated with the user to be configured parsed from the first configuration request. If the user to be configured is not found, the user to be configured is selected in the audit configuration data table first, and then the associated user audit configuration information is added. In this way, the object to be configured and the user audit configuration information associated with the user to be configured parsed from the first configuration request are added to the audit configuration data table in the form of a mapping relationship.

[0136] If the object to be configured includes a service to be configured, the service to be configured is used as the search dimension to search for and retrieve the service to be configured in the audit configuration data table of the database. If the service to be configured is found, the original user audit configuration information of the service to be configured in the audit configuration data table is modified according to the user audit configuration information associated with the service to be configured parsed from the first configuration request. If the service to be configured is not found, the service to be configured is selected in the audit configuration data table first, and then the associated user audit configuration information is added. In this way, the service to be configured and the user audit configuration information associated with the service to be configured parsed from the first configuration request are added to the audit configuration data table in the form of a mapping relationship.

[0137] The method provided in this embodiment offers a flexible configuration approach based on configuration requests. It allows for configuration of the audit configuration data table from multiple dimensions, such as user-level audit control and audit type control for specific businesses. This ensures both user-level and business-type audits, thereby guaranteeing the accuracy of interactions from multiple dimensions without affecting the normal use by non-auditing users or the execution logic of other businesses. It offers high flexibility and can better meet various interaction needs. Furthermore, it can be updated in real time as requirements change, allowing for modifications to the audit method and content at any time. It is simple and fast, has few dependent components, requires no additional development costs, and only requires configuration changes during updates, without modifying commands, thus improving the stability and real-time performance of interactions.

[0138] In some embodiments, step 104, which involves obtaining multiple inference models corresponding to the target task based on the plug-in / plug-out state and the service type, includes:

[0139] If the plug-in / plug-out status is determined to be enabled, the target audit type is determined in the audit configuration data table of the database according to the service type.

[0140] Based on the target audit type, multiple inference models are obtained from the business configuration data table in the database.

[0141] Figure 7 This is a schematic diagram of the process for obtaining the target review type provided in this embodiment; such as Figure 7 and Figure 3 As shown, the steps for obtaining the inference model include:

[0142] When the plug-in / plug-out state is enabled, i.e., when the human-computer interactive audit strategy is determined, the audit type determination module is called again. Based on the business type, the target audit type is determined, such as by using the "getBizConfigList(applicationId)" command. Then, based on the target audit type, multiple different inference models are called, such as by calling getBizConfigValidListAndOccupyByKeys(List <integer>The `applicationIdList` directive code retrieves a list of inference models; then, it performs target business processing based on multiple different inference models, obtaining multiple processing results. These processing results can exist in various forms, such as a list. <integer>The type exists.

[0143] like Figure 7 As shown, the specific steps for determining the target audit type are as follows:

[0144] Using the business type of the target business as the search dimension, the system searches for the target review type in the review configuration data table of the database to check whether a corresponding target review type is configured for the business type of the target business. If no corresponding target review type is configured, the search result is returned as empty, and the plug-in status of the user review module is updated to the disabled state. Instead, an interactive autonomous review strategy is adopted, that is, using intent information to directly obtain multiple inference models corresponding to the target business for inference, review, and configuration.

[0145] If a corresponding target review type is configured, the target review type is returned, and the user review module is kept in the enabled state. Multiple inference models are then retrieved from the business configuration data table in the database.

[0146] It should be noted that the relationship between business types and review types in the database's review configuration data table can be obtained by administrators selecting the corresponding business type and configuring the review type for that business type in the user configuration terminal, based on the business application of the Chat system. For example, the business type of acrostic poems will only have the review type set for acrostic poems, while the business type of ordinary dialogue will need to have review types set for dialogue, drawing, poetry, etc.

[0147] The method provided in this embodiment, when the user review module is enabled, can determine the target review type based on the business type and select an appropriate inference model for the review task. By flexibly selecting different inference models, the accuracy and efficiency of the review can be improved, adapting to different types of review needs, thereby enhancing the accuracy and applicability of the interaction.

[0148] In some embodiments, step 105, which involves reviewing and configuring multiple processing results based on the plugging / unplugging status to obtain response information corresponding to the dialogue request, and pushing the response information to the user requesting end, includes:

[0149] When the plugging / unplugging state is enabled, a target audit task is generated based on multiple processing results;

[0150] Store the target review task in the review queue;

[0151] The target review task is read from the review queue according to the trigger time and the storage order of the target review task.

[0152] Upon receiving the target review task, information to be reviewed is generated based on the target review task and the dialogue request;

[0153] The information to be reviewed is pushed to the user configuration terminal via a server push event connection with the user configuration terminal;

[0154] Obtain a second configuration request; the second configuration request includes a request entered when reviewing and configuring multiple processing results based on the information to be reviewed displayed on the user configuration terminal;

[0155] Based on the second configuration request, obtain the target configuration operation;

[0156] Based on the target configuration operation, configure multiple processing results to obtain response information corresponding to the dialogue request;

[0157] The response information corresponding to the dialogue request is pushed to the user requesting client through a server push event connection with the user requesting client.

[0158] Server Push Event Connection (SSE) is a web technology used to enable servers to push data to clients in real time. Based on the HTTP (Hypertext Transfer Protocol), it allows servers to send data to clients in the form of event streams. Clients can receive the data pushed by the server in real time by establishing a persistent HTTP connection and listening to the event stream.

[0159] Figure 8 This is one of the flowcharts illustrating SSE interaction provided in this embodiment; such as Figure 9 This is the second flowchart illustrating the SSE interaction provided in this embodiment, as follows: Figure 3 , Figure 8 and Figure 9 As shown, for system interaction, the SSE method is adopted. For page display to users and administrators, the server-side push method of the interactive device is used to push data to the interface in real time. The specific steps include the following:

[0160] When the plug-in / plug-out state is enabled, i.e., when the human-computer interactive review strategy is determined, a target review task is generated based on multiple processing results and stored in the review queue. This storage can be achieved using the "saveVerifyPool(chatId)" command. Then, the scheduled task module is called to loop through the latest review tasks in the review list to read the target review task from the review queue.

[0161] After reading the target review task, the available SSE information is obtained from the SSECacheMap (Server-Sent Events Cache Map). The SSECacheMap is a connection pool used to store the currently available connection information. If the administrator is reviewing, the current connection pool is empty. At this time, the information to be reviewed generated based on the target review task and the dialogue request can be directly pushed to the user configuration terminal via SSE.

[0162] For example, the following command can be used to push information pending review:

[0163] "sseEmmitter.send(SseEmitter.event().name("message").id(uniqueId).data(result))".

[0164] After logging into the user configuration client, the administrator waits for the pending review information pushed by the interactive device. Upon receiving the pending review information, clicking to open the message directly leads to the review page, which displays the user's input dialogue request and multiple processing results returned by the inference model. The administrator can then input operation commands (such as HTTP PUT requests, chat, verify, and submit) to select, modify, or submit multiple processing results, thereby generating a second configuration request.

[0165] After receiving the second configuration request, the interactive device configures multiple processing results, obtains response information corresponding to the dialogue request, stores the response information corresponding to the dialogue request in the database, and pushes the response information to the user requesting end in the form of SSE for display.

[0166] The method provided in this embodiment starts the human-computer interaction review channel when the plug-in state is enabled. Through interaction, different types of content are reviewed interactively, thereby ensuring that the response information fed back to the user request end is more reasonable and accurate. In addition, the interaction with the user configuration end and the user request end is carried out using SSE. Each request uses only one connection, which improves the real-time performance of the interaction response, reduces the pressure on the server end, improves the interaction response rate, and improves the availability and stability of the server.

[0167] In some embodiments, step 105 involves reviewing and configuring multiple processing results based on the plugging / unplugging status to obtain response information corresponding to the dialogue request, including...

[0168] When the user review module is in a disabled state, multiple processing results are reviewed and configured according to a preset review algorithm to obtain the response information corresponding to the dialogue request.

[0169] like Figure 3 As shown, when the user review module is in a disabled state, multiple inference models need to be directly invoked based on the intent information to process the target task and obtain multiple processing results.

[0170] Subsequently, the local preset review algorithm is loaded to automatically review and configure multiple processing results, and obtain the response information corresponding to the dialogue request. Then, the response information corresponding to the dialogue request is stored in the database and pushed to the user requesting end in the form of SSE for display.

[0171] The method provided in this embodiment automatically switches to the machine-autonomous review channel when the plug-in / plug-out state is disabled, thereby enabling automatic review of different types of content and ensuring that the response information fed back to the user request end is more reasonable and accurate.

[0172] The interactive device provided by the present invention will be described below. The interactive device described below can be referred to in correspondence with the interactive method described above.

[0173] Figure 10 This is a schematic diagram of the structure of the interactive device provided in this embodiment; as shown below. Figure 10 As shown, the device includes a front-end 1001 and a server; the server includes a processing module 1002, an audit verification module 1003, an audit type determination module 1004, an inference module 1005, and an audit push module 1006.

[0174] The front end 1001 is used to obtain dialogue requests; the dialogue requests include processing requests for target tasks input by the user request terminal;

[0175] The processing module 1002 is used to obtain the target user information corresponding to the user request terminal and the business type of the target task according to the dialogue request;

[0176] The audit verification module 1003 is used to determine the plug-in / plug-out status of the user audit module based on the target user information and the business type.

[0177] The audit type determination module 1004 is used to obtain multiple inference models corresponding to the target task based on the plug-in / plug-out status and the business type.

[0178] The reasoning module 1005 is used to process the target task based on multiple reasoning models to obtain multiple processing results;

[0179] The review and push module 1006 is used to review and configure multiple processing results according to the plug-in / plug-out status, obtain response information corresponding to the dialogue request, and push the response information to the user request terminal.

[0180] The interactive device provided in this embodiment obtains the target user information and the business type of the target task corresponding to the user request terminal through a dialogue request. Based on the target user information and business type, it determines the plug-in / plug-out status of the user review module. This allows for the adaptive acquisition of multiple inference models and the review and configuration of multiple processing results based on the plug-in / plug-out status of the user review module. By adaptively performing human-computer interactive review configuration or machine autonomous review configuration based on the inference results of multiple inference models, the final response result is obtained. This not only improves the completeness of the interaction but also enhances the accuracy and flexibility of the interaction.

[0181] In some embodiments, this embodiment also provides an interactive system, such as Figure 2 The interactive system includes an interactive device as provided in the above embodiments, at least one user request terminal, at least one user configuration terminal, and a database; the interactive device establishes communication with each of the user request terminals and each of the user configuration terminals respectively through server push event connection.

[0182] The interactive system provided in this embodiment obtains the target user information and the business type of the target task corresponding to the user requesting end through a dialogue request. Based on the target user information and business type, it determines the plug-in / plug-out status of the user review module. This allows for the adaptive acquisition of multiple inference models and the review and configuration of multiple processing results based on the plug-in / plug-out status of the user review module. By adaptively performing human-computer interactive review configuration or machine autonomous review configuration based on the inference results of multiple inference models, the final response result is obtained. This not only improves the completeness of the interaction but also enhances the accuracy and flexibility of the interaction.

[0183] Figure 11 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 11 As shown, the electronic device may include a processor 1101, a communications interface 1102, a memory 1103, and a communication bus 1104, wherein the processor 1101, the communications interface 1102, and the memory 1103 communicate with each other via the communication bus 1104. The processor 1101 can call logical instructions in the memory 1103 to execute an interaction method, which includes: acquiring a dialogue request; the dialogue request including a processing request for a target task input by a user request terminal; acquiring target user information corresponding to the user request terminal and the service type of the target task based on the dialogue request; determining the plug-in / plug-out status of a user review module based on the target user information and the service type; acquiring multiple inference models corresponding to the target task based on the plug-in / plug-out status and the service type; processing the target task based on the multiple inference models to obtain multiple processing results; reviewing and configuring the multiple processing results based on the plug-in / plug-out status to obtain response information corresponding to the dialogue request, and pushing the response information to the user request terminal.

[0184] Furthermore, the logical instructions in the aforementioned memory 1103 can be implemented as software functional modules and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program instructions, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0185] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the interaction methods provided by the above methods. The method includes: obtaining a dialogue request; the dialogue request includes a processing request for a target task input by a user request terminal; obtaining target user information corresponding to the user request terminal and the business type of the target task according to the dialogue request; determining the plug-in / plug-out state of a user review module according to the target user information and the business type; obtaining multiple inference models corresponding to the target task according to the plug-in / plug-out state and the business type; processing the target task based on the multiple inference models to obtain multiple processing results; reviewing and configuring the multiple processing results according to the plug-in / plug-out state to obtain response information corresponding to the dialogue request, and pushing the response information to the user request terminal.

[0186] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the interactive methods provided by the above methods. The method includes: acquiring a dialogue request; the dialogue request including a processing request for a target task input by a user requesting terminal; acquiring target user information corresponding to the user requesting terminal and the business type of the target task based on the dialogue request; determining the plug-in / plug-out state of a user review module based on the target user information and the business type; acquiring multiple inference models corresponding to the target task based on the plug-in / plug-out state and the business type; processing the target task based on the multiple inference models to obtain multiple processing results; reviewing and configuring the multiple processing results based on the plug-in / plug-out state to obtain response information corresponding to the dialogue request, and pushing the response information to the user requesting terminal.

[0187] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0188] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0189] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.< / integer> < / integer>

Claims

1. An interaction method, characterized in that, include: Obtain the dialogue request; The dialogue request includes a processing request for the target task input by the user request terminal; Based on the dialogue request, obtain the target user information corresponding to the user requesting end and the business type of the target task; Based on the target user information and the business type, determine the plug-in / plug-out status of the user review module; Based on the plug-in / plug-out status and the service type, multiple inference models corresponding to the target task are obtained, and the target task is processed based on the multiple inference models to obtain multiple processing results; Based on the plugging / unplugging status, multiple processing results are reviewed and configured to obtain response information corresponding to the dialogue request, and the response information is pushed to the user requesting end; The steps for determining the insertion / removal status include: In the database's audit configuration data table, find the user audit configuration information associated with the target user information; If user review configuration information associated with the target user information is found, it is determined whether user review configuration information is configured for the user corresponding to the target user information, and whether user review configuration information is configured for the business type, in order to determine the plug-in / plug-out status; the review configuration data table in the database is pre-configured with multiple sets of user information and user review configuration information in a mapping relationship; the mapping relationship between user information and user review configuration information is configured according to the user importance level and the business type importance level.

2. The interaction method according to claim 1, characterized in that, The step of obtaining the target user information corresponding to the user requesting end and the business type of the target task according to the dialogue request includes: The dialog request is format-converted to obtain a converted dialog request. Based on the converted dialogue request, obtain the target user information and the intent information of the dialogue request; Based on the intent information, the business type is obtained by matching it in the business information table of the database.

3. The interaction method according to claim 1, characterized in that, The step of determining the plug-in / plug-out status of the user review module based on the target user information and the business type includes: In the database's audit configuration data table, find the user audit configuration information associated with the target user information; If user review configuration information associated with the target user information is found, then the user review configuration information is searched for to find first user review authorization information associated with the target user information and second user review authorization information associated with the business type. If it is determined that a first user review authorization information associated with the target user information and a second user review authorization information associated with the business type are found, the plug-in / plug-out status of the user review module is updated to the enabled status.

4. The interaction method according to claim 3, characterized in that, The method further includes: Obtain a first configuration request; the first configuration request includes a configuration request for the audit configuration data table input through the user configuration terminal; Based on the first configuration request, obtain the object to be configured and the user approval configuration information associated with the object to be configured; the object to be configured includes the user to be configured and / or the service to be configured. Locate the object to be configured in the audit configuration data table; If the object to be configured is not found, the object to be configured and the user review configuration information associated with the object to be configured are added to the review configuration data table in the form of a mapping relationship.

5. The interaction method according to any one of claims 1-4, characterized in that, The step of obtaining multiple inference models corresponding to the target task based on the plug-in / plug-out status and the service type includes: If the plug-in / plug-out status is determined to be enabled, the target audit type is determined in the audit configuration data table of the database according to the service type. Based on the target audit type, multiple inference models are obtained from the business configuration data table in the database.

6. The interaction method according to any one of claims 1-4, characterized in that, The step of reviewing and configuring multiple processing results based on the plugging / unplugging status, obtaining response information corresponding to the dialogue request, and pushing the response information to the user requesting end includes: When the plugging / unplugging state is enabled, a target audit task is generated based on multiple processing results; Store the target review task in the review queue; The target review task is read from the review queue according to the trigger time and the storage order of the target review task. Upon receiving the target review task, information to be reviewed is generated based on the target review task and the dialogue request; The information to be reviewed is pushed to the user configuration terminal via a server push event connection with the user configuration terminal; Obtain a second configuration request; the second configuration request includes a request entered when reviewing and configuring multiple processing results based on the information to be reviewed displayed on the user configuration terminal; Based on the second configuration request, obtain the target configuration operation; Based on the target configuration operation, configure multiple processing results to obtain response information corresponding to the dialogue request; The response information corresponding to the dialogue request is pushed to the user requesting client through a server push event connection with the user requesting client.

7. The interaction method according to any one of claims 1-4, characterized in that, The process involves reviewing and configuring multiple processing results based on the plugging / unplugging status to obtain response information corresponding to the dialogue request, including... When the user review module is in a disabled state, multiple processing results are reviewed and configured according to a preset review algorithm to obtain the response information corresponding to the dialogue request.

8. An interactive device, characterized in that, Including front-end and server-side; The server includes a processing module, an audit verification module, an audit type determination module, an inference module, and an audit push module; The front end is used to obtain dialogue requests; the dialogue requests include processing requests for target tasks input by the user request terminal; The processing module is used to obtain the target user information corresponding to the user requesting end and the business type of the target task according to the dialogue request. The audit verification module is used to determine the plug-in / plug-out status of the user audit module based on the target user information and the business type. The audit type determination module is used to obtain multiple inference models corresponding to the target task based on the plug-in / plug-out status and the business type. The reasoning module is used to process the target task based on multiple reasoning models to obtain multiple processing results; The review and push module is used to review and configure multiple processing results according to the plug-in / plug-out status, obtain response information corresponding to the dialogue request, and push the response information to the user request terminal; The audit and verification module is also used for: In the database's audit configuration data table, search for user audit configuration information associated with the target user information; if user audit configuration information associated with the target user information is found, determine whether user audit configuration information is configured for the user corresponding to the target user information, and whether user audit configuration information is configured for the business type, to determine the plug-in / plug-out status; the database's audit configuration data table is pre-configured with multiple sets of user information and user audit configuration information in a mapping relationship; the mapping relationship between user information and user audit configuration information is configured according to the user importance level and the business type importance level.

9. An interactive system, characterized in that, Includes the interactive device as described in claim 8, at least one user request terminal, at least one user configuration terminal, and a database; The interactive device establishes communication with each of the user request terminals and each of the user configuration terminals through server push event connections.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the interaction method as described in any one of claims 1 to 7.

11. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the interaction method as described in any one of claims 1 to 7.

Images