Data storage method and device based on large model, storage medium, equipment and product

Abstract

The invention relates to the technical field of data storage, and particularly provides a data storage method and device based on a large model, a storage medium, equipment and a product, and the method can comprise the steps: obtaining text content returned by a target large model based on call information sent by a call center; if it is confirmed that the text content is the first return content of the target large model, the text content is stored in a database; and sending construction data corresponding to the text content to the call center, so that the call center executes a call task according to the construction data. According to some embodiments of the invention, complete storage of large model data can be realized, and the accuracy of data storage is guaranteed.

Description

Technical Field

[0001] This application relates to the technical field of data storage. Specifically, it relates to a method, device, storage medium, equipment, and product for data storage based on a large model. Background Art

[0002] With the development of intelligent customer service technology and the emergence of large model technology, large model technology has been widely applied to the field of intelligent telemarketing due to its excellent capabilities.

[0003] Currently, a large model can output reply content that matches the user's question. However, when the call center plays the text corresponding to the reply content to the user, the user may interrupt its playback process for some reason, resulting in the dialogue system being unable to completely store the reply content and causing the loss of background stored data.

[0004] Therefore, how to provide a technical solution for a complete method of data storage based on a large model has become a technical problem that urgently needs to be solved. Summary of the Invention

[0005] Some embodiments of this application aim to provide a method, device, storage medium, equipment, and product for data storage based on a large model. Through the technical solutions of the embodiments of this application, even when the user interrupts during the dialogue process, the integrity of the dialogue data storage can be guaranteed, and the risk of dialogue data loss can be avoided.

[0006] In a first aspect, some embodiments of this application provide a method for data storage based on a large model, including: obtaining the text content returned by a target large model based on call information sent by a call center; if it is confirmed that the text content is the first returned content of the target large model, storing the text content in a database; and sending construction data corresponding to the text content to the call center so that the call center can execute a call task according to the construction data.

[0007] Some embodiments of this application store the text content in the database after obtaining the text content returned by the target large model and confirming that it is the first returned content, and then send it to the call center. By storing the first returned text content, the embodiments of this application guarantee the integrity of the dialogue data storage even when the user interrupts during the subsequent dialogue process, and avoid the risk of dialogue data loss.

[0008] In some embodiments, the call information includes: user dialogue record, model version identifier, whether to add an interruption label, and whether to add a transfer to human operator script.

[0009] Some embodiments of this application can ensure the accuracy of the generated text content by transmitting the requirements of the call center through call information.

[0010] In some embodiments, before obtaining the text content returned by the target large model based on the call information sent by the call center, the method further includes: receiving the model version identifier sent by the call center, where the model version identifier includes a workspace ID and a conversation ID; and searching for the target large model that matches the model version identifier.

[0011] Some embodiments of the present application search for the corresponding target large model through the model version identifier sent by the call center to generate the conversation text in the call task, thereby achieving accurate processing of the call task.

[0012] In some embodiments, before sending the constructed data corresponding to the text content to the call center, the method further includes: adding constructed content to the text content to obtain the constructed data, where the constructed content includes at least one of the text label corresponding to the text content, the recording number, the transfer-to-human script, and the interruption label; whether the transfer-to-human script and the interruption label are added to the constructed content is determined based on the call information; the interruption label indicates whether the text content is allowed to be interrupted by the user during playback; and the recording number is used to search for the corresponding recording file later.

[0013] Some embodiments of the present application obtain constructed data by adding corresponding constructed content to the text content, enabling the call center to handle special requirements such as user transfer to human or interruption according to the constructed content, and enabling subsequent classification and search of conversation data through the text label and the recording number.

[0014] In some embodiments, after sending the constructed data corresponding to the text content to the call center, the method further includes: receiving the request information sent by the call center, where the request information is generated after the voice signal corresponding to the text content is interrupted during playback; obtaining the latest text content generated by the target large model based on the request information; and sending the latest constructed data corresponding to the latest text content to the call center, so that the call center can play the latest voice signal corresponding to the latest text content.

[0015] Some embodiments of the present application re-request the target large model to generate the latest text content after the user interruption and then play it to the user, thereby realizing timely update of the conversation content and improving the user experience.

[0016] In some embodiments, the method further includes: generating an alarm message and pushing it to the technician when it is confirmed that the text content returned by the target large model has not been received within a preset period.

[0017] Some embodiments of the present application monitor the target large model by confirming whether text content is received within a preset time period, enabling technicians to detect anomalies in a timely manner and ensuring the orderly progress of call tasks.

[0018] In a second aspect, some embodiments of the present application provide a device for data storage based on a large model, including: a return module for obtaining text content returned by the target large model based on call information sent by a call center; a storage module for storing the text content in a database if it is confirmed that the text content is the first returned content of the target large model; and a sending module for sending construction data corresponding to the text content to the call center so that the call center can execute call tasks according to the construction data.

[0019] In a third aspect, some embodiments of the present application provide a computer-readable storage medium having a computer program stored thereon, and when the program is executed by a processor, the method described in any embodiment of the first aspect can be implemented.

[0020] In a fourth aspect, some embodiments of the present application provide an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the program, the method described in any embodiment of the first aspect can be implemented.

[0021] In a fifth aspect, some embodiments of the present application provide a computer program product, the computer program product including a computer program, and when the computer program is executed by a processor, the method described in any embodiment of the first aspect can be implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] To more clearly illustrate the technical solutions of some embodiments of the present application, the drawings required for some embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0023] Figure 1 It is a system diagram of data storage based on a large model provided by some embodiments of the present application;

[0024] Figure 2 It is one of the method flowcharts of data storage based on a large model provided by some embodiments of the present application;

[0025] Figure 3 It is another method flowchart of data storage based on a large model provided by some embodiments of the present application;

[0026] Figure 4 Block diagram of a device for data storage based on a large model provided for some embodiments of the present application;

[0027] Figure 5 Schematic diagram of an electronic device provided for some embodiments of the present application. Detailed implementation manners

[0028] Next, the technical solutions in some embodiments of the present application will be described in conjunction with the accompanying drawings in some embodiments of the present application.

[0029] It should be noted that: Similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present application, the terms "first", "second", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0030] In the related art, in the field of telesales, during the process of the intelligent customer service terminal communicating with the user, normally, the large model of the current round of the intelligent customer service terminal will store the content fed back by the large model after the broadcast is completed; when the large model of the current round is broadcasting the voice content and is interrupted by the user, the content fed back by the large model is not updated to the database Redis. Therefore, the result returned by the large model of the previous round is obtained from Redis. In another case, after the user interrupts, only the content before the interruption is stored in the database, for example, only half a sentence. The above situations will all lead to the incompleteness of the storage of the conversation content, and complete data content cannot be obtained when it is necessary to review the conversation content or analyze the conversation content later.

[0031] In view of this, some embodiments of the present application provide a method for data storage based on a large model. After obtaining the text content returned by the target large model based on the call information, by determining whether the target large model returns this text content for the first time, if it is determined that it is the first return, the text content is stored; then the text content is constructed to obtain the constructed data and sent to the call center. Some embodiments of the present application first determine whether to store the text content before sending it to the call center, which can ensure that even if the broadcast is interrupted by the user later, the complete conversation content will be stored, providing complete data for subsequent comparison content analysis.

[0032] Next, the overall composition structure of the system for data storage based on a large model provided by some embodiments of the present application will be described by way of example. Figure 1

[0033] Figure 1 ​​As shown in the figure, some embodiments of the present application provide a system for data storage based on a large model. The system for data storage based on a large model includes: a user terminal 100, a call center 200, and a dialogue server 300. Among them, the dialogue server 300 includes a dialogue system and a large model applied to various business types. The business types may include telesales and adding WeChat in the marketing field, etc. For each large model corresponding to a business type, a workspaceID (as a specific example of a workspace ID) and a dialogueID (as a specific example of a dialogue ID) are set. Through these two IDs, a unique large model dialogue process can be determined. Each user determines a call through a unique sessionID. The call center 200 will create a new sessionID for each user and temporarily save the user's dialogue information (or referred to as dialogue data, dialogue content, etc.). Specifically, the reply text returned by the large model of the dialogue server 300 is converted into a voice signal by the call center 200 and sent to the user terminal 100. The user of the user terminal 100 transmits the voice signal to the call center 200 through a phone call, and the call center 200 converts the voice signal into text and requests the large model of the dialogue server 300.

[0034] In some embodiments of the present application, for the situation that the text content returned by the large model-based dialogue system of the above-mentioned dialogue server 300 is prone to timeout, the storage function of the model is separately split from the original function and a new port is opened. And when setting the load balancing for multiple servers of the dialogue server 300, the scheduling algorithm of weighted least connections (WLC) is selected. WLC distributes the requests of the call center 200 among multiple servers, so that the load of each server is allocated according to its weight. Suppose there are three servers A, B, and C (as a specific example of the dialogue server 300), with weights of 4, 2, and 1 respectively, and the current call connection numbers are 3, 2, and 1 respectively. Server A: 3 / 4 = 0.75, Server B: 2 / 2 = 1.00, Server C: 1 / 1 = 1.00. Select the server A with the smallest load index to process the new request and update its connection number to 4.

[0035] In some embodiments of the present application, the user terminal can be various types of mobile terminals (such as smart phones, iPads, etc.). The embodiments of the present application do not make specific limitations here.

[0036] The following combines the attached Figure 2 Exemplarily expounds the implementation process of data storage based on a large model executed by the dialogue server 300 provided by some embodiments of the present application.

[0037] Please refer to the attached Figure 2 , Figure 2A method flowchart for data storage based on a large model provided for some embodiments of the present application. The method for data storage based on a large model may include:

[0038] S210. Obtain the text content returned by the target large model based on the call information sent by the call center. The call information includes: user conversation record, model version identifier, whether to add an interruption label, and whether to add a transfer-to-human script.

[0039] For example, in some embodiments of the present application, when the call center 200 executes a call task, it needs to send call information to the dialogue system of the dialogue server 300 so that the dialogue system can provide the call information to the target large model. The target large model generates corresponding text content based on the user conversation record and feeds it back to the dialogue system. The content included in the call information can be set according to the actual application scenario, and the embodiments of the present application are not limited thereto.

[0040] In some embodiments of the present application, before executing S210, the method for data storage based on a large model may further include: receiving the model version identifier sent by the call center, where the model version identifier includes a workspace ID and a dialogue ID; and searching for the target large model that matches the model version identifier.

[0041] For example, in some embodiments of the present application, when the call center 200 executes a call task, it first determines the workspace ID (as a specific example of the workspace ID) and dialogue ID (as a specific example of the dialogue ID) of the target large model for executing the current call task based on the business type of the call task, and sends them to the dialogue system. After receiving the workspace ID and dialogue ID, the dialogue system searches for the corresponding large model version information from multiple large models, and then requests the target large model corresponding to the large model version information. Because the target large model and the user conversation require high real-time performance, a streaming return of the dialogue is required.

[0042] S220. If it is confirmed that the text content is the first returned content of the target large model, store the text content in the database.

[0043] For example, in some embodiments of the present application, the dialogue system can determine whether the text content is the first return of the target large model. If so, it is stored in the database Redis; otherwise, it is not stored. This can ensure the integrity of the storage of the text content returned by the target large model. Even if other problems occur during the broadcast from the call center 200 to the user terminal 100 later, it will not affect the storage of the output content of the target large model, providing a reliable basis for subsequent data analysis and use.

[0044] In some embodiments of the present application, before executing S230, the method for data storage based on a large model may further include: adding constructed content to the text content to obtain the constructed data, wherein the constructed content includes: a text tag corresponding to the text content, a recording number, at least one of the conversion to manual speech and the interruption tag; whether to add the conversion to manual speech and the interruption tag to the constructed content is determined based on the call information; the interruption tag indicates whether the user is allowed to interrupt when broadcasting the text content; the recording number is used for later searching for the corresponding recording file.

[0045] For example, in some embodiments of the present application, after receiving the text content of the target large model returned by streaming, the dialogue system adds the construction content inside it to obtain the construction data. Specifically, the construction content may include: adding text tags, such as user questions and sales skills tags; adding recording numbers, the program will search for the corresponding recording file according to the recording number; judging whether to transfer to a person, if the call information contains a transfer to manual service, then construct the transfer words and recording (as a specific example of the transfer to manual words); judging whether the text content can support interruption during the broadcasting process. By default, the words text (that is, the text content) supports interruption. For example, when the user says words such as complaint type, the robot of the call center 200 stops broadcasting and responds to the user's question; if the setting does not support interruption, the robot will not stop talking, and the voice corresponding to the text content needs to be broadcasted before responding to the user's words. For example, when verifying the user's identity, the robot speaks very briefly (Hello, hello, are you Mr.\Ms. XX?), and it can be set to not support interruption at this time. Whether to transfer to manual and the interruption tag setting can be set according to the actual application scenario, and the embodiment of the present application is not specifically limited here.

[0046] In some embodiments of the present application, the method for storing data based on a large model may further include: generating an alarm message and pushing it to a technician when confirming that the text content returned by the target large model is not received within a preset period of time.

[0047] For example, in some embodiments of the present application, the dialogue system can monitor the time for the target large model to return text content to avoid the target large model returning too slowly, resulting in the same prompt content in the two rounds of requesting large models. If the target large model returns for more than a preset period of time (for example, 1s), an alarm message needs to be sent to the alarm group, and the technician can be informed in advance whether the concurrency is insufficient and the system situation needs to be checked to ensure that the call task can proceed normally.

[0048] S230: Sending construction data corresponding to the text content to the call center, so that the call center can perform the call task according to the construction data.

[0049] For example, in some embodiments of the present application, the dialogue system streams the constructed data that has been completed to the call center 200. The call center 200 may convert the text content in the constructed data into voice information and broadcast it to the user of the client 100.

[0050] In some embodiments of the present application, after executing S230, the method for data storage based on a large model may further include: receiving request information sent by the call center, where the request information is generated after the voice signal corresponding to the text content is interrupted during broadcasting; obtaining the latest text content generated by the target large model based on the request information; and sending the latest constructed data corresponding to the latest text content to the call center, so that the call center can broadcast the latest voice signal corresponding to the latest text content.

[0051] For example, in some embodiments of the present application, after the call center 200 is interrupted during the process of broadcasting the recording corresponding to the voice signal to the client 100, the call center 200 will stop the recording broadcast and send request information to the dialogue server 300, request the interface of the target large model, and wait for the target large model to return the latest result for the current user's question (as a specific example of the latest text content). After that, after the call center 200 receives the latest constructed data, it will broadcast the latest voice signal corresponding to the latest text content, instead of continuing to broadcast the interrupted recording.

[0052] The following Figure 3 exemplarily elaborates on the specific process of data storage based on a large model provided by some embodiments of the present application.

[0053] Please refer to the Figure 3 , Figure 3 which is a flowchart of a method for data storage based on a large model provided by some embodiments of the present application.

[0054] The following exemplarily elaborates on the above process.

[0055] S310, the call center sends a call request message to the dialogue server.

[0056] Among them, the call request message is the call message mentioned above.

[0057] S320, the dialogue system of the dialogue server locates the corresponding target large model through the model version identifier in the call message.

[0058] S330, the dialogue system of the dialogue server receives the text content returned by the target large model based on the call request message.

[0059] S340, determine whether it is the first return. If so, execute S341; otherwise, execute S350.

[0060] S341, storing the text content in a database.

[0061] S350: The dialogue system constructs the text to obtain construction data.

[0062] S360, the dialogue system sends the construction data to the call center.

[0063] S370, the call center converts the text content in the constructed data into a voice signal.

[0064] S380: The call center broadcasts the recording corresponding to the voice signal to the user terminal.

[0065] It should be noted that the specific implementation process of S310 to S380 can refer to the method embodiments provided above, and detailed description is appropriately omitted here to avoid repetition. Through the above embodiments of the present application, the integrity of the conversation data storage can be guaranteed without duplication, and the reliability is high.

[0066] Please refer to Figure 4 , Figure 4 The block diagram of the composition of the device for storing data based on a large model provided by some embodiments of the present application is shown. It should be understood that the device for storing data based on a large model corresponds to the above method embodiment and can execute each step involved in the above method embodiment. The specific functions of the device for storing data based on a large model can be found in the description above. To avoid repetition, the detailed description is appropriately omitted here.

[0067] Figure 4 The device for storing data based on a big model includes at least one software function module that can be stored in a memory in the form of software or firmware or solidified in the device for storing data based on a big model. The device for storing data based on a big model includes: a return module 410, used to obtain the text content returned by the target big model based on the call information sent by the call center; a storage module 420, used to confirm that the text content is the first return content of the target big model, and then store the text content in a database; a sending module 430, used to send construction data corresponding to the text content to the call center, so that the call center can perform call tasks according to the construction data.

[0068] In some embodiments of the present application, the call information includes: user conversation record, model version identifier, whether to add an interruption tag, and whether to add a manual conversion script.

[0069] In some embodiments of the present application, before the return module 410, the apparatus for data storage based on a large model includes: a lookup module (not shown in the figure) configured to: receive the model version identifier sent by the call center, where the model version identifier includes a workspace ID and a conversation ID; and look up the target large model that matches the model version identifier.

[0070] In some embodiments of the present application, before the sending module 430, the apparatus for data storage based on a large model includes: a construction module (not shown in the figure) configured to: add construction content to the text content to obtain the construction data, where the construction content includes at least one of: a text label corresponding to the text content, a recording number, the manual transfer script, and the interruption label; whether to add the manual transfer script and the interruption label in the construction content is determined based on the call information; the interruption label indicates whether the text content is allowed to be interrupted by the user during playback; and the recording number is used to later find the corresponding recording file.

[0071] In some embodiments of the present application, after the sending module 430, the apparatus for data storage based on a large model includes: an update module (not shown in the figure) configured to: receive the request information sent by the call center, where the request information is generated after the voice signal corresponding to the text content is interrupted during playback; obtain the latest text content generated by the target large model based on the request information; and send the latest construction data corresponding to the latest text content to the call center, so that the call center can play the latest voice signal corresponding to the latest text content.

[0072] In some embodiments of the present application, the apparatus for data storage based on a large model includes: an alarm module (not shown in the figure) configured to: generate an alarm message and push it to the technician in the case of confirming that the text content returned by the target large model has not been received within a preset period.

[0073] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the above-described apparatus can refer to the corresponding process in the foregoing method, and will not be elaborated herein too much.

[0074] Some embodiments of the present application also provide a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the operations corresponding to any of the methods provided in the above embodiments of the above method can be implemented.

[0075] Some embodiments of the present application also provide a computer program product, which includes a computer program. When the computer program is executed by a processor, it can implement the operations of the corresponding methods in any of the above methods provided in the above embodiments.

[0076] As Figure 5 shown, some embodiments of the present application provide an electronic device 500, which includes: a memory 510, a processor 520, and a computer program stored on the memory 510 and executable on the processor 520. When the processor 520 reads the program from the memory 510 through a bus 530 and executes the program, it can implement the methods in any of the above embodiments.

[0077] The processor 520 can process digital signals and can include various computing architectures. For example, a complex instruction set computer architecture, a reduced instruction set computer architecture, or an architecture that implements a combination of multiple instruction sets. In some examples, the processor 520 can be a microprocessor.

[0078] The memory 510 can be used to store instructions executed by the processor 520 or data related to the execution of the instructions. These instructions and / or data can include code for implementing some or all of the functions of one or more modules described in the embodiments of the present application. The processor 520 in the embodiments of the present disclosure can be used to execute the instructions in the memory 510 to implement the methods shown above. The memory 510 includes dynamic random access memory, static random access memory, flash memory, optical memory, or other memories well known to those skilled in the art.

[0079] The above description is only for the embodiments of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numerals and letters indicate similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0080] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, and all of them should be covered by the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

[0081] Note that, in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an ..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

Claims

1. A method for data storage based on large models, characterized in that, include: Obtaining text content returned by the target large model based on the call information sent by the call center; Confirming that the text content is the first return content of the target large model, then storing the text content in a database; The construction data corresponding to the text content is sent to the call center so that the call center can perform the call task according to the construction data.

2. The method according to claim 1, wherein The call information includes: user conversation record, model version identifier, whether to add an interruption tag, and whether to add a manual conversion script.

3. The method according to claim 2, wherein Before acquiring the text content returned by the target macro model based on the call information sent by the call center, the method further includes: receiving the model version identifier sent by the call center, wherein the model version identifier includes a workspace ID and a conversation ID; The target large model matching the model version identifier is searched.

4. The method according to claim 2 or 3, characterized in that, Before sending the construction data corresponding to the text content to the call center, the method further includes: Adding construction content to the text content to obtain the construction data, wherein the construction content includes: a text tag corresponding to the text content, a recording number, at least one of the conversion to manual speech and the interruption tag; whether to add the conversion to manual speech and the interruption tag to the construction content is determined based on the call information; the interruption tag indicates whether the user is allowed to interrupt when broadcasting the text content; the recording number is used for later searching for the corresponding recording file.

5. The method according to any one of claims 1 to 3, characterized in that After sending the construction data corresponding to the text content to the call center, the method further includes: Receiving a request message sent by the call center, wherein the request message is generated after a voice signal corresponding to the text content is interrupted when broadcasting; Acquire the latest text content generated by the target macro model based on the request information; The latest structure data corresponding to the latest text content is sent to the call center so that the call center can broadcast the latest voice signal corresponding to the latest text content.

6. The method according to any one of claims 1 to 3, characterized in that The method further includes: generating an alarm message and pushing it to a technician when confirming that the text content returned by the target large model is not received within a preset period of time.

7. A device for data storage based on a large model, characterized in that, include: A return module, used to obtain the text content returned by the target large model based on the call information sent by the call center; A storage module, used for confirming that the text content is the first return content of the target large model, and then storing the text content in a database; The sending module is used to send the construction data corresponding to the text content to the call center, so that the call center can perform the call task according to the construction data.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program executes the method according to any one of claims 1 to 6 when executed by a processor.

9. An electronic device, characterized in that, The method comprises a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the computer program executes the method according to any one of claims 1 to 6 when being run by the processor.

10. A computer program product, characterized in that, The computer program product described above includes a computer program, wherein when the computer program is run by a processor, it executes the method described in any one of claims 1-6.

Images