Request data processing method and device, computer equipment, readable storage medium and program product
By receiving user requests in the interactive interface and determining the target module, the problem of insufficient adaptability and business relevance of traditional large models is solved, and efficient business processing is achieved.
Patent Information
- Application Number
- CN202411727266.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-28
- Publication Date
- 2025-02-11
AI Technical Summary
When traditional big models are applied in different subjects and business scenarios, the adaptability and business correlation are insufficient, resulting in inefficient business processing.
By receiving input parameter requests for user identification and module selection requests in the interactive interface, the target module (such as the proxy management module, the private domain knowledge management module and the multimodal data management module) are determined, and the requests are processed through these modules to obtain the target data. Finally, the target data is sent to the interactive interface.
A customized large model is realized to meet the actual needs of different entities and improve the efficiency of business processing.
Smart Images
Figure CN119415211A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and in particular to a request data processing method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art
[0002] With the continuous development of artificial intelligence technology, large model technology is increasingly used in various industries.
[0003] In traditional technologies, different entities develop large models that are strongly related to their own businesses. Since different entities and different business scenarios have different requirements for large models, when entities use large models that are not highly adaptable to the entity or have poor relevance to the business for business processing, the business processing efficiency is often low. Summary of the invention
[0004] Based on this, it is necessary to provide a request data processing method, apparatus, computer equipment, computer-readable storage medium and computer program product that can improve business processing efficiency in response to the above technical problems.
[0005] In a first aspect, the present application provides a request data processing method, the method comprising:
[0006] Receiving a request through an interactive interface, the request including a first request and a second request; the first request represents an input parameter request corresponding to a user identifier; the second request represents a module selection request corresponding to the user identifier; the interactive interface indicates an interface for interacting with a large model;
[0007] Determine a corresponding target module according to the second request; the target module includes at least one of an agent management module, a private domain knowledge management module and a multimodal data management module; wherein the agent management module is used to respond to the first request through an automated task; the private domain knowledge management module is used to respond to the first request through document information; and the multimodal data management module is used to respond to the first request through multimodal data;
[0008] Processing the first request through the target module to obtain target data;
[0009] The target data is sent to the interactive interface.
[0010] In one embodiment, before receiving the request through the interactive interface, the method further includes:
[0011] Receiving document information uploaded by the user ID;
[0012] Parsing the document information and establishing a document index corresponding to the document information;
[0013] Determine each knowledge base based on the document information and the corresponding document index;
[0014] Storing each of the knowledge bases in an initial private domain knowledge management module;
[0015] The initial private domain knowledge management module is tested to obtain a private domain knowledge management module.
[0016] In one embodiment, the testing of the initial private domain knowledge management module to obtain the private domain knowledge management module includes:
[0017] receiving a test request, and processing the test request through the initial private domain knowledge management module to obtain test result data;
[0018] When the test result data does not meet the result similarity threshold, adjusting the block of the document index in the initial private domain knowledge management module, and using the adjusted initial private domain knowledge management module as the private domain knowledge management module;
[0019] When the test result data meets the result similarity threshold, the initial private domain knowledge management module is used as the private domain knowledge management module.
[0020] In one embodiment, the target data includes at least one of an output parameter, document information corresponding to a document index, and target multimodal data; and the processing of the first request by the target module to obtain the target data includes at least one of the following:
[0021] Determining, by means of a logic engine in the proxy management module, a target plug-in for executing the first request;
[0022] Extracting input parameters from the first request, and calling the target plug-in to process the input parameters to obtain output parameters;
[0023] Extracting the input parameter from the first request, and determining, by the private domain knowledge management module, document information corresponding to the document index according to the input parameter;
[0024] The input parameter is extracted from the first request, and the target multimodal data is determined according to the input parameter by the multimodal data management module; wherein the target multimodal data includes at least one of audio data, video data and picture data.
[0025] In one embodiment, before sending the target data to the interactive interface, the method includes:
[0026] Calculating a similarity value between at least one of the output parameter, the document information corresponding to the document index, and the target multimodal data and the first request;
[0027] Obtaining a similarity value with the largest value from each of the similarity values;
[0028] The data corresponding to the similarity value with the largest numerical value is taken as the target data.
[0029] In one embodiment, before receiving the request through the interactive interface, the method further includes:
[0030] Receiving initial multimodal data uploaded by the user identifier, wherein the initial multimodal data includes at least one of audio data, video data, and picture data;
[0031] Preprocessing the initial multimodal data to obtain multimodal data;
[0032] The multimodal data is stored in the multimodal data management module.
[0033] In a second aspect, the present application further provides a request data processing device, the device comprising:
[0034] A first receiving module is used to receive a request through an interactive interface, wherein the request includes a first request and a second request; the first request represents an input parameter request corresponding to a user identifier; the second request represents a module selection request corresponding to the user identifier; the interactive interface indicates an interface for interacting with a large model;
[0035] a determination module, configured to determine a corresponding target module according to the second request; the target module includes at least one of an agent management module, a private domain knowledge management module, and a multimodal data management module; wherein the agent management module is configured to respond to the first request through an automated task; the private domain knowledge management module is configured to respond to the first request through document information; and the multimodal data management module is configured to respond to the first request through multimodal data;
[0036] A processing module, configured to process the first request through the target module to obtain target data;
[0037] A sending module is used to send the target data to the interactive interface.
[0038] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.
[0039] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.
[0040] In a fifth aspect, the present application also provides a computer program product, including a computer program, which implements the steps of the above method when executed by a processor.
[0041] The above-mentioned request data processing method, device, computer equipment, computer-readable storage medium and computer program product, first, receive a first request of an input parameter request corresponding to a user identifier and a second request of a module selection request corresponding to a user identifier on an interactive interface; the target module includes at least one of an agent management module, a private domain knowledge management module and a multimodal data management module; the agent management module, the private domain knowledge management module and the multimodal data management module are all processing modules constructed according to the internal data of a certain subject; secondly, determine the corresponding target module according to the second request; process the first request through the target module to obtain the target data; finally, send the target data to the interactive interface. On the one hand, a customized large model is obtained by building and deploying the agent management module, the private domain knowledge management module and the multimodal data management module, so that the model can meet the actual needs of different subjects. On the other hand, the first request is processed by selecting the agent management module, the private domain knowledge management module and the multimodal data management module constructed using the internal data according to the second request, that is, the business request is processed through a highly customized large model, which improves the efficiency of processing business. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0043] FIG1 is an application environment diagram of a request data processing method in one embodiment;
[0044] FIG2 is a schematic diagram of a flow chart of a method for processing request data in one embodiment;
[0045] FIG3 is a schematic diagram of an interactive interface in one embodiment;
[0046] FIG4 is a schematic diagram of a process for determining a private domain knowledge management module in one embodiment;
[0047] FIG5 is a schematic diagram of a process of obtaining a private domain knowledge management module according to a test in one embodiment;
[0048] FIG6 is a schematic diagram of a process of processing a first request to obtain target data in one embodiment;
[0049] FIG7 is a schematic diagram of a process for determining target data in one embodiment;
[0050] FIG8 is a schematic diagram of a process for determining a multimodal data management module in one embodiment;
[0051] FIG9 is a structural block diagram of a request data processing device in one embodiment;
[0052] FIG. 10 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0053] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0054] The request data processing method provided in the embodiment of the present application can be applied in the application environment shown in Figure 1. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The server receives a request through an interactive interface, and the request includes a first request and a second request; the first request represents an input parameter request corresponding to the user identifier; the second request represents a module selection request corresponding to the user identifier; the interactive interface indicates an interface for interacting with the large model; the corresponding target module is determined according to the second request; the target module includes at least one of an agent management module, a private domain knowledge management module and a multimodal data management module; wherein the agent management module is used to respond to the first request through an automated task; the private domain knowledge management module is used to respond to the first request through document information; the multimodal data management module is used to respond to the first request through multimodal data; the first request is processed through the target module to obtain the target data; and the target data is sent to the interactive interface. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The server 104 may be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.
[0055] The request data processing method provided in the embodiment of the present application can also be applied to a terminal. The processor of the terminal receives a request through an interactive interface, and the request includes a first request and a second request; the first request represents an input parameter request corresponding to a user identifier; the second request represents a module selection request corresponding to the user identifier; the interactive interface indicates an interface for interacting with a large model; the corresponding target module is determined according to the second request; the target module includes at least one of an agent management module, a private domain knowledge management module, and a multimodal data management module; wherein the agent management module is used to respond to the first request through an automated task; the private domain knowledge management module is used to respond to the first request through document information; the multimodal data management module is used to respond to the first request through multimodal data; the first request is processed through the target module to obtain target data; and the target data is sent to the interactive interface. Among them, the terminal can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, Internet of Things devices, and portable wearable devices.
[0056] In an exemplary embodiment, as shown in FIG2 , a request data processing method is provided, which is described by taking the method applied to a terminal as an example, and includes the following steps S202 to S208 . Among them:
[0057] Step S202: receiving a request through an interactive interface, where the request includes a first request and a second request.
[0058] Among them, the first request represents the input parameter request corresponding to the user identifier, that is, it represents the user-triggered input parameter request corresponding to the user identifier; the second request represents the module selection request corresponding to the user identifier, that is, it represents the user-triggered module selection request corresponding to the user identifier; the interactive interface indicates the interface for interacting with the large model.
[0059] The interactive interface includes an input parameter area, a module selection tab, including a private domain knowledge management module, a multimodal data management module and an agent management module, and a display area for target data returned after processing the first request, as shown in FIG3 .
[0060] Optionally, for example, an administrator A of a certain enterprise issues a second request through an interactive interface, by clicking the tab of the multimodal data management module on the interactive interface; the second request is issued; and the terminal deployed with the large model receives the second request. An administrator A of a certain enterprise issues a first request through an interactive interface, by entering the first request in the input parameter area on the interactive interface, and clicking the send button after the input is completed, which is to issue the first request; and the terminal deployed with the large model receives the first request. Among them, the terminal can receive the first request first, and then receive the second request after receiving the first request; the terminal can also receive the second request first, and then receive the first request after receiving the second request.
[0061] For example, the first request may be "where is the venue for the xx meeting held", "what to eat for lunch today", "what is the weather like today", etc. The second request represents a module selection request corresponding to the user identifier, such as selecting at least one module from the private domain knowledge management module, the multimodal data management module, and the proxy management module to process the first request. For example, the first request "where is the venue for the xx meeting held" is processed by the multimodal data management module; for another example, the first request "where is the venue for the xx meeting held" is processed by the multimodal data management module and the private domain knowledge management module.
[0062] Step S204, determining a corresponding target module according to the second request; the target module includes at least one of an agent management module, a private domain knowledge management module and a multimodal data management module.
[0063] Among them, the agent management module is used to respond to the first request through an automated task; the private domain knowledge management module is used to respond to the first request through document information; and the multimodal data management module is used to respond to the first request through multimodal data.
[0064] Optionally, the terminal determines the target module according to the module selection tab clicked by the user in the interactive interface. For example, in the scenario of retrieving a certain document, the user clicks on the private domain knowledge management module in the interactive interface, and the private domain knowledge management module is determined as the processing module. In the scenario of processing some complex and repetitive tasks, such as weather query and order processing, the user can also click on the agent management module in the interactive interface to determine the agent management module as the processing module.
[0065] Optionally, if the user clicks on the private domain knowledge management module and the multimodal data management module in the interactive interface according to different scenarios, it is determined that the private domain knowledge management module and the multimodal data management module are both processing modules. The user can also click on the private domain knowledge management module, the multimodal data management module and the agent management module in the interactive interface, and all three modules are used as processing modules.
[0066] Step S206: Process the first request through the target module to obtain target data.
[0067] Optionally, the terminal processes the first request through a target module determined according to the second request. For example, if the target module is a private domain knowledge management module, the private domain knowledge management module processes the first request for large model parameter debugging, obtains target data, and determines whether the target data meets the conditions. If not, the large model parameter debugging is changed. For another example, the target module is an agent management module, which receives the first request input by the user. The agent management module intelligently matches and calls the corresponding plug-in for processing according to the first request, and outputs the target data. For another example, the target module is a multimodal data management module, which receives the first request input by the user. The multimodal data management module obtains the corresponding target data from the stored multimodal data according to the first request, and outputs the target data.
[0068] Step S208, sending the target data to the interactive interface.
[0069] Optionally, the terminal sends the target data to a target data display area of an interaction interface between the user and the large model.
[0070] In the above-mentioned request data processing method, first, a first request of an input parameter request corresponding to a user identifier and a second request of a module selection request corresponding to a user identifier are received on an interactive interface; the target module includes at least one of an agent management module, a private domain knowledge management module and a multimodal data management module; the agent management module, the private domain knowledge management module and the multimodal data management module are processing modules constructed based on internal enterprise data; secondly, the corresponding target module is determined according to the second request; the first request is processed by the target module to obtain the target data; finally, the target data is sent to the interactive interface. On the one hand, a customized large model is obtained by building and deploying the agent management module, the private domain knowledge management module and the multimodal data management module, so that the model can meet the actual needs of the enterprise. On the other hand, the first request is processed by selecting the agent management module, the private domain knowledge management module and the multimodal data management module constructed using the internal enterprise data according to the second request, that is, the business request is processed by a highly customized large model, and the target data in response to the first request is sent to the interactive interface, thereby improving the efficiency of processing the business.
[0071] In an exemplary embodiment, the method for determining the private domain knowledge management module, as shown in FIG4 , includes steps S402 to S410 before receiving a request through the interactive interface. Among them:
[0072] Step S402: receiving document information uploaded by the user identification.
[0073] Among them, document information refers to the documents and materials within a certain entity, such as the meeting minutes of a certain enterprise and the contract information of a certain enterprise.
[0074] Optionally, the terminal receives document information uploaded by the user identification, such as receiving a PDF format document uploaded by enterprise administrator A, or receiving a doc format document uploaded by enterprise administrator B.
[0075] Step S404: parse the document information and create a document index corresponding to the document information.
[0076] Optionally, the terminal parses the document information through a processor, and establishes a document index corresponding to each document information, so as to retrieve and query the document more quickly.
[0077] Step S406: determine each knowledge base based on the document information and the corresponding document index.
[0078] Optionally, the terminal establishes knowledge bases of different topics, such as meeting minutes topics, contract topics, etc., based on the parsed document information and corresponding document indexes through a processor.
[0079] Step S408: storing each knowledge base into the initial private domain knowledge management module.
[0080] Optionally, the terminal stores each knowledge base into an initial private domain knowledge management module through a processor, so as to manage document information in the knowledge base.
[0081] Step S410, testing the initial private domain knowledge management module to obtain the private domain knowledge management module.
[0082] Optionally, the terminal tests the initial private domain knowledge management module through a processor to obtain the private domain knowledge management module.
[0083] In this embodiment, by pre-storing documents and materials within a subject into an initial private domain knowledge management module and then testing the initial private domain knowledge management module, a private domain knowledge management module is obtained to ensure the accuracy and efficiency of the private domain knowledge management module in processing requests.
[0084] In an exemplary embodiment, as shown in FIG5 , the initial private domain knowledge management module is tested to obtain the private domain knowledge management module, including steps S502 to S506. In which:
[0085] Step S502: receiving a test request, processing the test request through the initial private domain knowledge management module, and obtaining test result data.
[0086] Optionally, a test environment is set up in advance. In the test environment, the user sends a test request in the interactive interface, the terminal receives the request sent by the user in the interactive interface, and the processor of the terminal processes the test request by calling the initial private domain knowledge management module to obtain test result data.
[0087] Step S504: when the test result data does not meet the relevant result similarity threshold, the block of the document index in the initial private domain knowledge management module is adjusted, and the adjusted initial private domain knowledge management module is used as the private domain knowledge management module.
[0088] Optionally, since the initial private domain knowledge management module stores document information and corresponding document indexes, when the test result data does not meet the result similarity threshold, the processor of the terminal adjusts the blocking of the document index in the initial private domain knowledge management module and uses the adjusted initial private domain knowledge management module as the private domain knowledge management module.
[0089] Optionally, when the test result data does not meet the result similarity threshold, the terminal can also adjust the randomness parameters for the private domain knowledge management module or the result similarity threshold for the private domain knowledge management module through an adjustment instruction issued by the user, and use the adjusted initial private domain knowledge management module as the private domain knowledge management module.
[0090] Step S506: when the test result data meets the result similarity threshold, the initial private domain knowledge management module is used as the private domain knowledge management module.
[0091] Optionally, when the test result data meets the result similarity threshold, the initial private domain knowledge management module is used as the private domain knowledge management module, indicating that the initial private domain knowledge management module meets the test requirements and can be used as the private domain knowledge management module to be finally used.
[0092] In this embodiment, the initial private domain knowledge management module is adjusted according to the test result data, and the private domain knowledge management module is adjusted to a module that meets the test requirements, thereby ensuring the correctness of subsequent processing by the private domain knowledge management module.
[0093] In an exemplary embodiment, as shown in FIG6 , the target data includes at least one of an output parameter, document information corresponding to a document index, and target multimodal data; the first request is processed by a target module to obtain the target data, including at least one of steps S602 to S606, wherein:
[0094] Step S602: determine the target plug-in for executing the first request through the logic engine in the proxy management module; extract input parameters from the first request, and call the target plug-in to process the input parameters to obtain output parameters.
[0095] The logic engine is used to match the corresponding plug-in for the request according to the plug-in matching rules.
[0096] Optionally, the terminal determines the target plug-in to execute the first request according to the plug-in matching rules through the logic engine in the proxy management module. The terminal extracts input parameters from the first request. For example, if the first request is "What to eat for lunch today", an input parameter "lunch" is extracted from the first request. For another example, if the first request is "What is the weather like today", an input parameter "weather" is extracted from the first request. For another example, if the first request is "What are the orders with a delivery date of xx / xx / xx", two input parameters "delivery date of xx / xx / xx" and "orders" are extracted. The terminal processes the input parameters by calling the target plug-in through the proxy management module determined according to the second request to obtain output parameters.
[0097] It should be noted that the first request may include multiple input parameters, and the specific number is not limited here.
[0098] Step S604: extract input parameters from the first request, and determine document information corresponding to the document index according to the input parameters through the private domain knowledge management module.
[0099] Optionally, the first request is “Please find the contract signed with XX Company”, the input parameter “XX Company Contract” is extracted from the first request, and the document information corresponding to the document index is determined according to the input parameter by the private domain knowledge management module determined according to the second request, and the document information corresponding to the document index is used as the target data.
[0100] Step S606: extract input parameters from the first request, and determine target multimodal data according to the input parameters through the multimodal data management module; wherein the target multimodal data includes at least one of audio data, video data and picture data.
[0101] Optionally, the first request is "where is the location of XX meeting", and the terminal extracts the input parameter "XX meeting address" from it, and determines the target multimodal data from the multimodal data according to the input parameter through the multimodal data management module, such as a map, or a video of a map.
[0102] In this embodiment, by selecting different processing modules to process the first request in different scenarios, the business processing efficiency is effectively improved.
[0103] Continuing with the above embodiment, as shown in FIG7 , before sending the target data to the interactive interface, steps S702 to S706 are included. Among them:
[0104] Step S702, calculating a similarity value between at least one of the output parameter, the document information corresponding to the document index, and the target multimodal data and the first request.
[0105] Optionally, if the second request selects the proxy management module and the private domain knowledge management module, the terminal calculates the similarity value between the output parameters, the document information corresponding to the document index and the first request, such as the similarity value between the output parameters and the first request is A; the similarity value between the document information corresponding to the document index and the first request is B.
[0106] Optionally, if the second request selects the proxy management module, the private domain knowledge management module, and the multimodal data management module, the similarity value between the output parameter, the document information corresponding to the document index, and the target multimodal data and the first request is calculated. For example, the similarity value between the output parameter and the first request is A; the similarity value between the document information corresponding to the document index and the first request is B; and the similarity value between the target multimodal data and the first request is C.
[0107] Step S704: Obtain the similarity value with the largest value from each similarity value.
[0108] Optionally, the terminal obtains a similarity value with a maximum value from AB, such as B, or obtains a similarity value with a maximum value from ABC, such as C.
[0109] Step S706: taking the data corresponding to the similarity value with the largest value as the target data.
[0110] Optionally, if the second request selects the proxy management module, the private domain knowledge management module, or the multimodal data management module, the document information corresponding to the document index is used as the target data. If the second request selects the proxy management module, the private domain knowledge management module, or the multimodal data management module, the target multimodal data is used as the target data.
[0111] Optionally, the terminal may also send the output parameter, the document information corresponding to the document index, and the target multimodal data to the user interface.
[0112] In this embodiment, by outputting similarity values between the results of each module and the first request and selecting the one with the largest similarity value as the target data, the accuracy of the target data can be improved.
[0113] In an exemplary embodiment, as shown in FIG8 , before receiving a request through the interactive interface, the method further includes steps S802 to S806 . Among them:
[0114] Step S802: receiving initial multimodal data uploaded by a user identifier.
[0115] The initial multimodal data includes at least one of audio data, video data and picture data.
[0116] Optionally, initial multimodal data, such as audio data and video data, uploaded by enterprise administrator A is received. For another example, image data uploaded by enterprise administrator A is received.
[0117] Step S804: preprocess the initial multimodal data to obtain multimodal data.
[0118] Optionally, the terminal preprocesses the initial multimodal data, such as image recognition, speech recognition, normalization, denoising, etc., to obtain multimodal data.
[0119] Step S806: store the multimodal data in a multimodal data management module.
[0120] Optionally, the terminal stores the multimodal data in the multimodal data management module. When the user's second request is to select the multimodal data management module, the multimodal data management module processes the first request, thereby determining target multimodal data matching the first request from the multimodal data.
[0121] Optionally, after storing the multimodal data in the multimodal data management module, a test environment is pre-built. Under the test environment, the user sends a test request in the interactive interface, the terminal receives the request sent by the user in the interactive interface, and the processor of the terminal processes the test request by calling the multimodal data management module to obtain test result data. When the test result data does not meet the relevant result similarity threshold, the processor of the terminal adjusts the randomness parameter of the multimodal data management module or the result similarity threshold for the multimodal data management module. When the test result data meets the relevant result similarity threshold, it indicates that the randomness parameter for the multimodal data management module or the result similarity threshold for the multimodal data management module meets the test requirements, and the multimodal data management module passes the test.
[0122] Similarly, for the proxy management module, multiple plug-ins that respond to requests are pre-deployed. After the multiple plug-ins that respond to requests are deployed, in the test environment, the user sends a test request in the interactive interface, the terminal receives the request sent by the user in the interactive interface, and the processor of the terminal processes the test request by calling the proxy management module to intelligently match the target plug-in to obtain the test result data. When the test result data does not meet the relevant result similarity threshold, the processor of the terminal adjusts the randomness parameters of the multimodal data management module, the result similarity threshold for the multimodal data management module, or the plug-in matching rules. When the test result data meets the relevant result similarity threshold, it indicates that the randomness parameters for the proxy management module, the result similarity threshold for the proxy management module, and the plug-in matching rules meet the test requirements, and the proxy management module passes the test.
[0123] In this embodiment, by pre-storing the multimodal data in the multimodal data management module in accordance with the requirements of the multimodal data management module, when the multimodal data management module is used to process the request later, the standard multimodal data is returned to the interactive interface to improve the accuracy of the multimodal data. Through the multimodal management module, the processing capability of multimodal data such as images and audio is added to enrich the form of the target data.
[0124] In an exemplary embodiment, it is divided into two phases, a deployment test phase and a use phase. The interactive interface includes an input parameter area, a module selection tab, including a private domain knowledge management module, a multimodal data management module and an agent management module, and a display area for the target data returned after processing the first request.
[0125] Deployment test phase: Users send document information through the interactive interface, and the terminal receives PDF format documents uploaded by enterprise administrator A and doc format documents uploaded by enterprise administrator B. The terminal parses the document information, establishes a knowledge base of different topics based on the document index corresponding to each document information, and stores it in the initial private domain knowledge management module.
[0126] A test environment is pre-built. Under the test environment, the user sends a test request in the interactive interface, the terminal receives the request sent by the user in the interactive interface, and the processor of the terminal processes the test request by calling the initial private domain knowledge management module to obtain the test result data. Since the document information and the corresponding document index are stored in the initial private domain knowledge management module, when the test result data does not meet the result similarity threshold, the processor of the terminal adjusts the block of the document index in the initial private domain knowledge management module, and uses the adjusted initial private domain knowledge management module as the private domain knowledge management module. Since the document information and the corresponding document index are stored in the initial private domain knowledge management module, when the test result data does not meet the result similarity threshold, the processor of the terminal adjusts the block of the document index in the initial private domain knowledge management module, and uses the adjusted initial private domain knowledge management module as the private domain knowledge management module. The terminal can also adjust the randomness parameters for the private domain knowledge management module or the result similarity threshold for the private domain knowledge management module through the adjustment instructions issued by the user, and use the adjusted initial private domain knowledge management module as the private domain knowledge management module.
[0127] Receive the initial multimodal data uploaded by enterprise administrator A, such as audio data and video data. Preprocess the initial multimodal data, such as image recognition, speech recognition, normalization, denoising, etc., to obtain multimodal data. The terminal stores the multimodal data in the multimodal data management module. When the user's second request is to select a multimodal data management module, the first request is processed by the multimodal data management module, so as to determine the target multimodal data that matches the first request from the multimodal data. In a test environment, the user sends a test request in the interactive interface, the terminal receives the request sent by the user in the interactive interface, and the processor of the terminal processes the test request by calling the multimodal data management module to obtain the test result data. When the test result data does not meet the relevant result similarity threshold, the processor of the terminal adjusts the randomness parameter of the multimodal data management module or the result similarity threshold for the multimodal data management module. When the test result data meets the relevant result similarity threshold, it indicates that the randomness parameter for the multimodal data management module or the result similarity threshold for the multimodal data management module meets the test requirements, and the multimodal data management module passes the test.
[0128] For the proxy management module, multiple plug-ins that respond to requests are pre-deployed. After the multiple plug-ins that respond to requests are deployed, in the test environment, the user sends a test request in the interactive interface, the terminal receives the request sent by the user in the interactive interface, and the processor of the terminal processes the test request by calling the proxy management module to intelligently match the target plug-in to obtain the test result data. When the test result data does not meet the relevant result similarity threshold, the processor of the terminal adjusts the randomness parameters of the multimodal data management module, the result similarity threshold for the multimodal data management module, or the plug-in matching rules. When the test result data meets the relevant result similarity threshold, it indicates that the randomness parameters for the proxy management module, the result similarity threshold for the proxy management module, and the plug-in matching rules meet the test requirements, and the proxy management module passes the test.
[0129] Use stage: The above modules have been tested. For example, the administrator A of a certain enterprise first issues the second request through the interactive interface. Clicking the tab of the multimodal data management module on the interactive interface is to issue the second request. The terminal with the large model deployed receives the second request. The administrator A of a certain enterprise then issues the first request through the interactive interface. Entering the first request in the input parameter area on the interactive interface and clicking the send button after entering is to issue the first request. The terminal with the large model deployed receives the first request.
[0130] The terminal selects the tab of the module clicked by the user in the interactive interface and determines the target module. For example, in the scenario of retrieving a certain document, if the user clicks the private domain knowledge management module in the interactive interface, the private domain knowledge management module is determined as the processing module. If the first request is "Please find the contract signed with XX Company", the input parameter "XX Company Contract" is extracted from the first request, and the document information corresponding to the document index is determined according to the input parameter by the private domain knowledge management module determined according to the second request, and the document information corresponding to the document index is used as the target data. The terminal sends the document information to the target data display area of the interactive interface between the user and the large model.
[0131] In the scenario of processing some complex and repetitive tasks, such as weather query and order processing, the user can also click on the proxy management module in the interactive interface to determine the proxy management module as the processing module. The user sends the first request and the second request through the display interface. The terminal receives the first request and the second request. The terminal determines the target plug-in to execute the first request according to the plug-in matching rules through the logic engine in the proxy management module. The first request is "What are the orders with a delivery date of xx / xx / xx?", from which two input parameters "delivery date of xx / xx / xx" and "order" are extracted. The target plug-in is called by the proxy management module determined according to the second request to process the input parameters and obtain the output parameters. The terminal sends the output parameters to the target data display area of the interactive interface between the user and the large model.
[0132] In the scenario of multimodal data processing, the user sends a first request and a second request through the display interface. The terminal receives the first request and the second request. If the first request is "Where is the location of the XX meeting", the second request is the request issued by the user clicking the proxy management module in the interactive interface, and the proxy management module is determined as the processing module according to the second request. The terminal extracts the input parameter "XX meeting address" from it, and determines the target multimodal data, such as a map, or a video of a map, from the multimodal data through the multimodal data management module according to the input parameter. The terminal sends the map or video to the target data display area of the interactive interface between the user and the large model.
[0133] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.
[0134] Based on the same inventive concept, the embodiment of the present application also provides a request data processing device for implementing the request data processing method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the one or more request data processing device embodiments provided below can refer to the limitations of the request data processing method above, and will not be repeated here.
[0135] In an exemplary embodiment, as shown in FIG. 9 , a request data processing device is provided, including: a first receiving module 901, a determining module 902, a processing module 903 and a sending module 904, wherein:
[0136] The first receiving module 901 is used to receive requests through an interactive interface, the requests including a first request and a second request; the first request represents an input parameter request corresponding to a user identifier; the second request represents a module selection request corresponding to the user identifier; the interactive interface indicates an interface for interacting with a large model.
[0137] Determination module 902 is used to determine the corresponding target module according to the second request; the target module includes at least one of an agent management module, a private domain knowledge management module and a multimodal data management module; wherein the agent management module is used to respond to the first request through an automated task; the private domain knowledge management module is used to respond to the first request through document information; and the multimodal data management module is used to respond to the first request through multimodal data.
[0138] The processing module 903 is used to process the first request through the target module to obtain target data.
[0139] The sending module 904 is used to send the target data to the interactive interface.
[0140] In an exemplary embodiment, a request data processing device includes:
[0141] The second receiving module is used to receive document information uploaded by the user identification.
[0142] The parsing module is used to parse the document information and establish a document index corresponding to the document information.
[0143] The knowledge base determination module is used to determine each knowledge base based on document information and corresponding document indexes.
[0144] The first storage module is used to store each knowledge base into the initial private domain knowledge management module.
[0145] The testing module is used to test the initial private domain knowledge management module to obtain the private domain knowledge management module.
[0146] In an exemplary embodiment, the test module includes:
[0147] The receiving unit is used to receive a test request, process the test request through the initial private domain knowledge management module, and obtain test result data.
[0148] The first determining unit is configured to adjust the block of the document index in the initial private domain knowledge management module when the test result data does not meet the relevant result similarity threshold, and use the adjusted initial private domain knowledge management module as the private domain knowledge management module.
[0149] The second determining unit is configured to use the initial private domain knowledge management module as the private domain knowledge management module when the test result data meets the result similarity threshold.
[0150] In an exemplary embodiment, the target data includes at least one of an output parameter, document information corresponding to a document index, and target multimodal data; the processing module 903 includes:
[0151] The first processing unit is used to determine the target plug-in for executing the first request through the logic engine in the proxy management module; extract input parameters from the first request, and call the target plug-in to process the input parameters to obtain output parameters.
[0152] The second processing unit is used to extract input parameters from the first request, and determine the document information corresponding to the document index according to the input parameters through the private domain knowledge management module.
[0153] The third processing unit is used to extract input parameters from the first request, and determine target multimodal data according to the input parameters through the multimodal data management module; wherein the target multimodal data includes at least one of audio data, video data and picture data.
[0154] In an exemplary embodiment, a request data processing device further includes:
[0155] A calculation module is used to calculate a similarity value between at least one of the output parameter, the document information corresponding to the document index, and the target multimodal data and the first request.
[0156] The acquisition module is used to obtain the similarity value with the largest value from each similarity value.
[0157] The target data determination module is used to take the data corresponding to the similarity value with the largest numerical value as the target data.
[0158] In an exemplary embodiment, a request data processing device further includes:
[0159] The third receiving module is used to receive initial multimodal data uploaded by the user identification, wherein the initial multimodal data includes at least one of audio data, video data and picture data.
[0160] The preprocessing module is used to preprocess the initial multimodal data to obtain multimodal data.
[0161] The second storage module is used to store the multimodal data in the multimodal data management module.
[0162] Each module in the above-mentioned request data processing device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.
[0163] In an exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be shown in FIG10. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store output parameters, document information corresponding to the document index and target multimodal data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a request data processing method is implemented.
[0164] Those skilled in the art will understand that the structure shown in FIG. 10 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.
[0165] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.
[0166] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.
[0167] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.
[0168] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.
[0169] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.
[0170] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the present application. It should be noted that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A request data processing method, characterized in that: The method comprises: Receiving a request through an interactive interface, the request including a first request and a second request; the first request represents an input parameter request corresponding to a user identifier; the second request represents a module selection request corresponding to the user identifier; the interactive interface indicates an interface for interacting with a large model; Determine a corresponding target module according to the second request; the target module includes at least one of an agent management module, a private domain knowledge management module and a multimodal data management module; wherein the agent management module is used to respond to the first request through an automated task; the private domain knowledge management module is used to respond to the first request through document information; and the multimodal data management module is used to respond to the first request through multimodal data; Processing the first request through the target module to obtain target data; The target data is sent to the interactive interface.
2. The method according to claim 1, characterized in that Before receiving the request through the interactive interface, the method further includes: Receiving document information uploaded by the user identification; Parsing the document information and establishing a document index corresponding to the document information; Determine each knowledge base based on the document information and the corresponding document index; Storing each of the knowledge bases in an initial private domain knowledge management module; The initial private domain knowledge management module is tested to obtain a private domain knowledge management module.
3. The method according to claim 2, characterized in that The testing of the initial private domain knowledge management module to obtain the private domain knowledge management module includes: receiving a test request, and processing the test request through the initial private domain knowledge management module to obtain test result data; When the test result data does not meet the result similarity threshold, adjusting the block of the document index in the initial private domain knowledge management module, and using the adjusted initial private domain knowledge management module as the private domain knowledge management module; When the test result data meets the result similarity threshold, the initial private domain knowledge management module is used as the private domain knowledge management module.
4. The method according to claim 1, characterized in that The target data includes at least one of an output parameter, document information corresponding to a document index, and target multimodal data; the target module processes the first request to obtain the target data, including at least one of the following: Determining, by means of a logic engine in the proxy management module, a target plug-in for executing the first request; Extracting input parameters from the first request, and calling the target plug-in to process the input parameters to obtain output parameters; Extracting the input parameter from the first request, and determining, by the private domain knowledge management module, document information corresponding to the document index according to the input parameter; The input parameter is extracted from the first request, and the target multimodal data is determined according to the input parameter by the multimodal data management module; wherein the target multimodal data includes at least one of audio data, video data and picture data.
5. The method according to claim 4, characterized in that Before sending the target data to the interactive interface, the method includes: Calculating a similarity value between at least one of the output parameter, the document information corresponding to the document index, and the target multimodal data and the first request; Obtaining a similarity value with the largest value from each of the similarity values; The data corresponding to the similarity value with the largest numerical value is taken as the target data.
6. The method according to claim 1, characterized in that Before receiving the request through the interactive interface, the method further includes: Receiving initial multimodal data uploaded by the user identifier, wherein the initial multimodal data includes at least one of audio data, video data, and picture data; Preprocessing the initial multimodal data to obtain multimodal data; The multimodal data is stored in the multimodal data management module.
7. A request data processing device, characterized in that: The device comprises: A first receiving module is used to receive a request through an interactive interface, wherein the request includes a first request and a second request; the first request represents an input parameter request corresponding to a user identifier; the second request represents a module selection request corresponding to the user identifier; the interactive interface indicates an interface for interacting with a large model; a determination module, configured to determine a corresponding target module according to the second request; the target module includes at least one of an agent management module, a private domain knowledge management module, and a multimodal data management module; wherein the agent management module is configured to respond to the first request through an automated task; the private domain knowledge management module is configured to respond to the first request through document information; and the multimodal data management module is configured to respond to the first request through multimodal data; A processing module, configured to process the first request through the target module to obtain target data; A sending module is used to send the target data to the interactive interface.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.