Information processing method, information processing device, computer device, and storage medium

Abstract

The application relates to an information processing method, an information processing device, computer equipment and a storage medium. The method comprises the following steps: obtaining an information processing request triggered by a user account; the information processing request carries a dialogue sentence of the user account, and the dialogue sentence represents a request intention of the information processing request; based on the dialogue sentence, a target data table is selected from a plurality of data tables stored in a plurality of databases; information data stored in the target data table matches the request intention; the request intention and attribute information of the target data table are input into a pre-trained large language model to output an operation instruction for the information processing request; the operation instruction is represented based on a structured operation language; and the operation instruction is executed to call a corresponding function interface to perform data processing on the information data in the target data table, so as to obtain a task processing result for the information processing request. The above method can improve the efficiency of information processing on the request intention of the user and the rationality of the generated operation instruction.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an information processing method, an information processing apparatus, a computer device, a computer-readable storage medium, and a computer program product. Background Technology

[0002] Large Language Models (LLMs, which are generative models), such as ChatGPT (Chat Generative Pre-trained Transformer, a chatbot program developed by OpenAI), can generate and execute many operational instructions in response to user requests for information processing (i.e., requests for tasks). For example, requests for processing application data related to a user's account can generate and execute a series of operational tasks.

[0003] However, the current methods for applying LLM to the real world to generate and execute operation instructions are not very convenient for users. Users still need to input multiple rounds of dialogue to get operation instructions that better match their intentions, which results in low efficiency in instruction generation and execution. Summary of the Invention

[0004] This disclosure provides an information processing method, an information processing apparatus, a computer device, a computer-readable storage medium, and a computer program product to at least solve the problems of low efficiency in generating and executing operation instructions in related technologies. The technical solution of this disclosure is as follows:

[0005] According to a first aspect of the present disclosure, an information processing method is provided, comprising:

[0006] Obtain information processing requests triggered by user accounts; the information processing requests carry dialogue statements of the user accounts, and the dialogue statements represent the request intent of the information processing requests.

[0007] Based on the dialogue statement, a target data table is selected from multiple data tables stored in multiple databases; the information data stored in the target data table matches the request intent.

[0008] The request intent and the attribute information of the target data table are input into a pre-trained large language model to output operation instructions for the information processing request; the operation instructions are based on structured operation language representations.

[0009] The operation instruction is executed to call the corresponding function interface to process the information data in the target data table and obtain the task processing result for the information processing request.

[0010] In one exemplary embodiment, the step of filtering a target data table from multiple data tables stored in multiple databases based on the dialogue statement includes:

[0011] The intent of the dialogue statement is identified to obtain the request intent for the dialogue statement;

[0012] Based on the request type of the request intent, candidate databases of the corresponding database type are selected from multiple databases; the request type of the request intent matches the database type of the candidate database.

[0013] The request intent is correlated and matched with the candidate data tables in the candidate database to determine the target data table that matches the request intent from the candidate database.

[0014] In an exemplary embodiment, the step of associating and matching the request intent with candidate data tables in the candidate database to determine a target data table that matches the request intent from the candidate database includes:

[0015] The request intent is converted into a first type of spatial vector, and the table creation statements of each candidate data table in the candidate database are converted into a second type of spatial vector; the table creation statements are used to express the data type of the information data stored in the candidate data tables.

[0016] The correlation degree is calculated between the first type of spatial vector and each of the second type of spatial vectors to obtain the correlation score for each of the candidate data tables;

[0017] Among the candidate data tables, the candidate data table with the highest correlation score is selected as the target data table.

[0018] In one exemplary embodiment, the number of request intents for the dialogue statement includes at least two;

[0019] The step of inputting the request intent and the attribute information of the target data table into a pre-trained large language model to output operation instructions for the information processing request includes:

[0020] Based on the attribute information of the target data table corresponding to each of the aforementioned request intentions, a data definition language is determined for each of the aforementioned target data tables; the data definition language is used to characterize the structure, field meanings, and relationships between fields of the target data table.

[0021] Based on the large language model, instructions are generated from the data definition language of each request intent and its corresponding target data table to obtain operation instructions for the information processing request.

[0022] In an exemplary embodiment, the step of generating instructions based on the data definition language of each request intent and its corresponding target data table according to the large language model to obtain operation instructions for the information processing request includes:

[0023] Based on the large language model, prompts are constructed for each of the request intentions and their corresponding target data tables to obtain operation prompt statements for each of the request intentions.

[0024] Based on the large language model, each of the operation prompt statements is converted into a structured operation statement to obtain operation instructions for each of the request intentions;

[0025] Based on the large language model, the operation instructions are arranged sequentially to obtain a corresponding set of operation instructions, and the set of operation instructions is used as the operation instructions for the information processing request.

[0026] In one exemplary embodiment, the information processing request also carries account information of the user account;

[0027] Before executing the operation instruction, the method further includes:

[0028] Obtain the index information of the target data table corresponding to the operation instruction;

[0029] The user account is subject to permission verification based on the index information and the account information; the permission verification is used to verify whether the user account has the functional permission to process data in the target data table.

[0030] If the permission verification passes, the operation instruction is executed.

[0031] In one exemplary embodiment, executing the operation instruction includes:

[0032] Identify the operation type of the operation instruction and the table type of the target data table; the table type includes offline type and real-time type;

[0033] When the table type is offline, the function interface corresponding to the operation type is called to perform offline data processing on the information data in the target data table.

[0034] When the table type is real-time, the function interface corresponding to the operation type is called to perform real-time data processing on the information data in the target data table.

[0035] In one exemplary embodiment, the step of calling the corresponding functional interface to process the information data in the target data table to obtain the task processing result for the information processing request includes:

[0036] If the permission verification passes, obtain the user's account number and password;

[0037] The account number and password are input into the function interface, so that the information data in the target data table can be processed sequentially through the function interface to perform data unlocking, data extraction and data calculation, so as to obtain the task processing result for the information processing request.

[0038] In one exemplary embodiment, before executing the operation instruction, the method further includes:

[0039] The operation instructions are subjected to dangerous syntax verification and dangerous syntax filtering to obtain the processed operation instructions;

[0040] The dangerous syntax verification process is used to verify whether there are abnormal functions and / or abnormal characters in the operation instruction;

[0041] The dangerous syntax filtering process is used to modify abnormal functions in the operation instructions into normal functions and / or remove abnormal characters in the operation instructions.

[0042] In one exemplary embodiment, after receiving the operation instruction for the information processing request, the method further includes:

[0043] Obtain the user account's optimization request for the operation instruction input; the optimization request carries the user account's optimization statement, the optimization statement representing the user account's optimization intention for each operation instruction corresponding to the request intention;

[0044] Based on the large language model, the data definition language of each optimization intent, request intent and its corresponding target data table is constructed to obtain optimized prompt statements for each request intent;

[0045] Based on the large language model, each of the optimized prompt statements is converted into a structured operation statement to obtain optimized instructions for each of the request intentions;

[0046] Based on the large language model, the optimized instructions are arranged sequentially to obtain the corresponding set of optimized instructions;

[0047] Each optimization instruction in the optimization instruction set is subjected to dangerous syntax verification and dangerous syntax filtering to obtain an optimization instruction for the information processing request.

[0048] According to a second aspect of the present disclosure, an information processing apparatus is provided, comprising:

[0049] The request acquisition unit is configured to execute an information processing request triggered by acquiring a user account; the information processing request carries a dialogue statement of the user account, and the dialogue statement represents the request intent of the information processing request.

[0050] The target data table unit is configured to perform a process based on the dialogue statement to filter out a target data table from multiple data tables stored in multiple databases; the information data stored in the target data table matches the request intent.

[0051] The instruction output unit is configured to input the request intent and the attribute information of the target data table into a pre-trained large language model to output an operation instruction for the information processing request; the operation instruction is based on a structured operation language representation.

[0052] The instruction execution unit is configured to execute the operation instruction to call the corresponding function interface to process the information data in the target data table and obtain the task processing result for the information processing request.

[0053] Thirdly, this application also provides a computer device, comprising:

[0054] processor;

[0055] Memory for storing the executable instructions of the processor;

[0056] The processor is configured to execute the executable instructions to implement the information processing method described above.

[0057] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium includes program data, which, when executed, implements the information processing method described above.

[0058] Fifthly, this application also provides a computer program product. The computer program product includes program instructions that, when executed, implement the information processing method described above.

[0059] The technical solutions provided by the embodiments of this disclosure have at least the following beneficial effects:

[0060] This method first obtains information processing requests triggered by user accounts; the information processing requests carry dialogue statements from the user accounts, which represent the request intent; based on the dialogue statements, a target data table is selected from multiple data tables stored in multiple databases; the information data stored in the target data table is matched with the request intent; the request intent and the attribute information of the target data table are input into a pre-trained large language model to output operation instructions for the information processing request; the operation instructions are based on structured operation language representations; the operation instructions are executed to call the corresponding functional interfaces to process the information data in the target data table, obtaining the task processing result for the information processing request. In this way, on the one hand, this solution determines the target data table that matches the user's request intent through the user's dialogue statements. Then, using the request intent and the target data table, it obtains and executes operation instructions for the information processing request, thereby optimizing the information processing flow, effectively improving the efficiency of information processing based on the user's request intent, and reducing the consumption of manpower and resources. On the other hand, unlike existing technologies, this solution uses a pre-trained language model to output a structured operation language that matches the attribute information of the user's request intent and the target data table. Data processing is then performed according to this structured operation language to obtain the processing result, thus optimizing the generation method of operation instructions, effectively improving the rationality of the generated operation instructions, and reducing the deviation between the final processing result and the user's actual request intent, resulting in a better user experience.

[0061] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0062] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0063] Figure 1 This is an application environment diagram illustrating an information processing method according to an exemplary embodiment.

[0064] Figure 2 This is a flowchart illustrating an information processing method according to an exemplary embodiment.

[0065] Figure 3 This is a flowchart illustrating a step of filtering a target data table according to an exemplary embodiment.

[0066] Figure 4 This is a flowchart illustrating the steps of generating operation instructions using a large language model according to an exemplary embodiment.

[0067] Figure 5 This is a flowchart illustrating the steps of executing an operation instruction according to an exemplary embodiment.

[0068] Figure 6 This is a flowchart illustrating a step for optimizing a prompt statement according to another exemplary embodiment.

[0069] Figure 7 This is a flowchart illustrating an information processing method according to another exemplary embodiment.

[0070] Figure 8 This is a block diagram illustrating a method for generating and optimizing operational tasks according to an exemplary embodiment.

[0071] Figure 9 This is a block diagram illustrating a method for performing an operation task according to an exemplary embodiment.

[0072] Figure 10 This is a block diagram of an information processing apparatus according to an exemplary embodiment.

[0073] Figure 11 This is a block diagram illustrating a computer device for information processing according to an exemplary embodiment. Detailed Implementation

[0074] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0075] The term "and / or" in the embodiments of this application refers to any and all possible combinations including one or more of the associated listed items. It should also be noted that, when used in this specification, "including / comprising" specifies the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, and / or components and / or groups thereof.

[0076] The terms "first," "second," etc., used in this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0077] Furthermore, although the terms "first," "second," etc., are used repeatedly in this application to describe various operations (or various elements or various applications or various instructions or various data), these operations (or elements or applications or instructions or data) should not be limited by these terms. These terms are only used to distinguish one operation (or element or application or instruction or data) from another operation (or element or application or instruction or data).

[0078] The information processing method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a communication network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be located in the cloud or on other network servers.

[0079] In some embodiments, reference Figure 1 First, server 104 obtains an information processing request triggered by a user account. This request carries a dialogue statement from the user account, representing the intent of the request. Then, based on the dialogue statement, server 104 filters out a target data table from multiple data tables stored in multiple databases. The information data stored in the target data table matches the request intent. Next, server 104 inputs the request intent and the attribute information of the target data table into a pre-trained large language model to output operation instructions for the information processing request. These operation instructions are based on a structured operation language representation. Finally, server 104 executes the operation instructions to call the corresponding functional interface to process the information data in the target data table, obtaining the task processing result for the information processing request.

[0080] In some embodiments, terminal 102 (such as a mobile terminal or a fixed terminal) can be implemented in various forms. Terminal 102 can be a mobile terminal, including mobile phones, smartphones, laptops, portable handheld devices, personal digital assistants (PDAs), tablet computers (PADs), etc., which can output operation instructions for information processing requests using a pre-trained large language model. Terminal 102 can also be a fixed terminal, including Automated Teller Machines (ATMs), automated multi-function machines, digital TVs, desktop computers, fixed-line computers, etc., which can output operation instructions for information processing requests using a pre-trained large language model.

[0081] Hereinafter, it is assumed that terminal 102 is a fixed terminal. However, those skilled in the art will understand that, if there are operations or elements specifically designed for mobile purposes, the construction according to the embodiments disclosed in this application can also be applied to mobile type terminal 102.

[0082] In some embodiments, the data processing component running on server 104 may load any of the various additional server applications and / or middleware applications being executed, such as HTTP (Hypertext Transfer Protocol), FTP (File Transfer Protocol), CGI (Common Gateway Interface), RDBMS (Relational Database Management System), etc.

[0083] In some embodiments, server 104 may be implemented using a standalone server or a server cluster consisting of multiple servers. Server 104 may be adapted to run one or more application services or software components that provide the terminal 102 described in the foregoing disclosure.

[0084] In some embodiments, the application service may include a service interface that provides the application service to the user (e.g., an operation interface for inputting dialogue statements, a display interface for displaying the target data table / operation instructions), and corresponding program services, etc. The software component may include, for example, an application program (SDK) or client (APP) that has the function of calling corresponding function interfaces to process information data in the target data table according to operation instructions.

[0085] In some embodiments, users can input corresponding code data or control parameters into the APP or client through the input device in the terminal 102 to execute the application services of the computer program in the server 104 and display the application services in the user interface.

[0086] In some embodiments, the operating system running on the APP or client may include various versions of Microsoft Windows®, Apple Macintosh® and / or Linux operating systems, various commercial or UNIX-like® operating systems (including but not limited to various GNU / Linux operating systems, Google Chrome® OS, etc.) and / or mobile operating systems such as iOS®, Windows® Phone, Android® OS, BlackBerry® OS, Palm® OS, and other online or offline operating systems, without specific limitations herein.

[0087] In some embodiments, such as Figure 2 As shown, an information processing method is provided, which is applied to... Figure 1 Taking server 104 as an example, the method includes the following steps:

[0088] Step S11: Obtain the information processing request triggered by the user account.

[0089] In one embodiment, the information processing request carries a dialog statement and account information of the user account.

[0090] Dialogue statements represent the intent to request information processing.

[0091] As an example, the user account first enters the dialogue statement "I want to know my live streaming account's live streaming revenue in the last seven days" on the user terminal. Then, the user terminal obtains the dialogue statement and the user account's account information (including account name, account type, etc.) to generate an information processing request.

[0092] Step S12: Based on the dialogue statements, select the target data table from multiple data tables stored in multiple databases.

[0093] The information stored in the target data table matches the request intent.

[0094] For example, if a user account's request intent is "to earn revenue from live streaming," then the information stored in the target data table needs to be live streaming data related to the user account; similarly, if a user account's request intent is "to recommend e-commerce products," then the information stored in the target data table needs to be e-commerce product information data related to the user account.

[0095] In one exemplary embodiment, see Figure 3 , Figure 3 This is a flowchart illustrating an embodiment of filtering a target data table in this application. Step S12, the process by which the server filters the target data table from multiple data tables stored in multiple databases, can be implemented in the following way:

[0096] Step S121: Perform intent recognition on the dialogue statement to obtain the request intent for the dialogue statement.

[0097] The number of request intents for the dialogue statement includes at least two.

[0098] In some embodiments, the server performs intent recognition on the dialogue statements based on a pre-trained intent recognition model to obtain the request intent for the dialogue statements.

[0099] Among them, the number of request intents for the dialogue statement includes at least two, that is, at least two user intents need to be completed in order to meet the user's needs.

[0100] For example, the intent recognition model analyzes the request "I want to know the live streaming revenue of my live streaming account in the last seven days" to determine the four request intents that need to be completed to meet the user's needs: "determine the number of live streaming sessions in the last seven days", "calculate the amount of gift revenue in each live streaming session", "calculate the amount of platform revenue in each live streaming session", and "calculate the total amount of revenue between gift revenue and platform revenue across all live streaming sessions".

[0101] Step S122: Based on the request type of the request intent, filter out candidate databases of the corresponding database type from multiple databases.

[0102] In one embodiment, the request type represents the library type of the database to be mapped to the request intent, that is, the request type of the request intent matches the library type of the candidate database.

[0103] The database type is the same as the table type of the data tables stored in the database, and all data tables stored in each database are of the same type.

[0104] As an example, a request intent needs to map to a database S of type A, so that the request type A' of the request intent matches type A, and all data tables stored in database S are of type A.

[0105] The database types must include at least three types: MySQL, Hive, and FlinkSQL. Therefore, the request types must also include databases of at least three types, corresponding to the request intent.

[0106] Step S123: Perform association matching between the request intent and the candidate data tables in the candidate database to determine the target data table that matches the request intent from the candidate database.

[0107] Specifically, the server associates and matches the request intent with the data content stored in each candidate data table in the candidate database to determine the target data table that matches the request intent from the candidate database.

[0108] As an example, a request intent is "to calculate the amount of gift revenue in each live stream session"; candidate data table S1 stores data on "platform revenue", candidate data table S2 stores data on "gift revenue", and candidate data table S3 stores data on "live stream time". The server calculates the degree of correlation between the request intent and candidate data table S1 (P1), the degree of correlation between the request intent and candidate data table S2 (P2), and the degree of correlation between the request intent and candidate data table S3 (P3), and the degree of correlation P2 > the degree of correlation P1 > the degree of correlation P3.

[0109] Step S13: Input the request intent and the attribute information of the target data table into the pre-trained large language model to output operation instructions for the information processing request.

[0110] The operation instructions are based on a structured operation language representation. For example, operation instructions are represented using SQL (Structured Query Language).

[0111] In one exemplary embodiment, see Figure 4 , Figure 4 This is a flowchart illustrating an embodiment of generating operation instructions for the large language model in this application. In step S13, the server inputs the request intent and attribute information of the target data table into the pre-trained large language model to output operation instructions for the information processing request. This can be implemented in the following way:

[0112] Step S131: Based on the attribute information of the target data table corresponding to each request intent, determine the data definition language of each target data table.

[0113] In one embodiment, the attribute information of the target data table includes the table name, table number, etc.

[0114] Data definition language (DDL) is used to characterize the structure of a target data table, the meaning of its fields, and the relationships between fields.

[0115] Step S132: Based on the large language model, generate instructions for each request intent and its corresponding target data table using the data definition language to obtain operation instructions for the information processing request.

[0116] Specifically, the large language model predicts instructions based on the data definition language of each request intent and the corresponding target data table to generate operation instructions for each request intent, thereby obtaining a set of operation instructions as operation instructions for information processing requests.

[0117] Step S14: Execute the operation instruction to call the corresponding function interface to process the information data in the target data table and obtain the task processing result for the information processing request.

[0118] In one embodiment, before executing the operation instruction, the server may further include: performing dangerous syntax verification and dangerous syntax filtering on the operation instruction to obtain the processed operation instruction.

[0119] The various operation instructions include multiple types of execution functions (including command execution functions, database operation functions, file operation functions, etc.) and operation characters. Dangerous syntax is composed of various illegal execution functions and operation characters, intended to be mixed into legitimate execution functions and operation characters.

[0120] Specifically, after the server generates each operation command, it needs to protect the operation commands against vulnerabilities such as dangerous syntax injection (e.g., SQL injection statements like drop table) to ensure the legitimacy and security of subsequent operations on the data table.

[0121] In one embodiment, the dangerous syntax verification process is used to verify whether there are abnormal functions and / or abnormal characters in the operation instruction; the dangerous syntax filtering process is used to modify the abnormal functions in the operation instruction to normal functions and / or remove the abnormal characters in the operation instruction.

[0122] In one exemplary embodiment, see Figure 5 , Figure 5 This is a flowchart illustrating an embodiment of the execution of operation instructions in this application. In step S13, the server calls the corresponding functional interface to process the information data in the target data table. This can be implemented in the following way:

[0123] Step S141: Identify the operation type of the operation instruction and the table type of the target data table.

[0124] The table types include offline and real-time types. For example, there are offline table types such as MySQL and Hive, as well as real-time table types such as FlinkSQL.

[0125] The operation type of the operation command is the SQL type, including query command type (SELECT), insert command type (INSERT), update command type (UPDATE), or delete command type (DELETE), etc.

[0126] Step S142: If the table type is offline, call the function interface corresponding to the operation type to perform offline data processing on the information data in the target data table.

[0127] As an example, when the target data table is of type MySQL or Hive, the corresponding first-type functional interface is called to process the information data in the target data table offline. The interface type of the first-type functional interface corresponds to the operation type of the identified operation command.

[0128] Step S143: If the table type is real-time, call the function interface corresponding to the operation type to perform real-time data processing on the information data in the target data table.

[0129] As an example, when the target data table is of type FlinkSQL, the corresponding second type of functional interface is called to process the information data in the target data table in real time. The interface type of the second type of functional interface corresponds to the operation type of the identification operation command.

[0130] In the aforementioned information processing process, the server obtains information processing requests triggered by user accounts; the information processing requests carry dialogue statements from the user accounts, which represent the request intent; based on the dialogue statements, a target data table is selected from multiple data tables stored in multiple databases; the information data stored in the target data table matches the request intent; the request intent and the attribute information of the target data table are input into a pre-trained large language model to output operation instructions for the information processing request; the operation instructions are based on structured operation language representations; the operation instructions are executed to call the corresponding functional interfaces to process the information data in the target data table, obtaining the task processing result for the information processing request. In this way, on the one hand, this solution determines the target data table that matches the user's request intent through the user's dialogue statements. Then, using the request intent and the target data table, it obtains and executes operation instructions for the information processing request, thereby optimizing the information processing flow, effectively improving the efficiency of information processing based on the user's request intent, and reducing the consumption of manpower and resources. On the other hand, unlike existing technologies, this solution uses a pre-trained language model to output a structured operation language that matches the attribute information of the user's request intent and the target data table. Data processing is then performed according to this structured operation language to obtain the processing result, thus optimizing the generation method of operation instructions, effectively improving the rationality of the generated operation instructions, and reducing the deviation between the final processing result and the user's actual request intent, resulting in a better user experience.

[0131] In an exemplary embodiment, step S123, which is the process by which the server performs association matching between the request intent and the candidate data tables in the candidate database to determine the target data table that matches the request intent from the candidate database, can be implemented in the following way:

[0132] Step 1: Convert the request intent into a first-class spatial vector, and convert the table creation statements of each candidate table in the candidate database into a second-class spatial vector.

[0133] The table creation statement is used to express the data type of the information data stored in the candidate data table.

[0134] Step 2: Calculate the correlation degree between the first type of spatial vector and each second type of spatial vector to obtain the correlation score for each candidate data table.

[0135] Step 3: Select the candidate data table with the highest correlation score from each candidate data table as the target data table.

[0136] As an example, a request intent is "to calculate the amount of gift revenue in each live stream session"; the table creation statement for candidate data table S1 indicates that candidate data table S1 stores information data on "platform revenue", the table creation statement for candidate data table S2 indicates that candidate data table S2 stores information data on "gift revenue", and the table creation statement for candidate data table S3 indicates that candidate data table S3 stores information data on "live stream time". The server calculates the correlation score P1 between the request intent and candidate data table S1, the correlation score P2 between the request intent and candidate data table S2, and the correlation score P3 between the request intent and candidate data table S3, where correlation score P2 > correlation score P1 > correlation score P3. Therefore, the server selects candidate data table S2 as the target data table.

[0137] In an exemplary embodiment, step S132, which is the process by which the server generates instructions based on the data definition language of each request intent and its corresponding target data table according to the large language model, to obtain operation instructions for the information processing request, can be implemented in the following ways:

[0138] Step 1: Based on the large language model, construct prompts for each request intent and its corresponding target data table to obtain operation prompt statements for each request intent.

[0139] Among them, the operation prompt statements are prompt statements for the operation tasks to be performed on the target data table in response to the request intention, and each task prompt is expressed in the form of natural language.

[0140] For example, the operation prompt for the request intent "determine the live broadcast sessions in the last seven days" is "extract the user account's live broadcast records from XX month XX day to XX month XX day from the target data table M1, and then determine the live broadcast sessions corresponding to each live broadcast record"; the operation prompt for the request intent "calculate the gift revenue amount in each live broadcast session" is "extract the gift revenue amount received by the user account in each live broadcast session from XX month XX day to XX month XX day from the target data table M2, and then calculate the total gift amount between the gift revenue amounts of each live broadcast session".

[0141] Step 2: Based on the large language model, each operation prompt statement is converted into a structured operation statement to obtain operation instructions for each request intent.

[0142] The SQL operation task is a structured executable program.

[0143] Step 3: Arrange the operation instructions sequentially based on the large language model to obtain the corresponding operation instruction set, and use the operation instruction set as the operation instructions for the information processing request.

[0144] As an example, the large model first converts the data to obtain the first operation instruction "to count the number of live broadcast sessions in the last seven days", the second operation instruction "to count the amount of gift revenue in each live broadcast session", the third operation instruction "to count the amount of platform revenue in each live broadcast session", and the fourth operation instruction "to count the total amount of revenue between gift revenue and platform revenue across all live broadcast sessions". Then, the first, second, third and fourth operation instructions are arranged in sequence to obtain the operation instructions for the information processing request.

[0145] In one exemplary embodiment, see Figure 6 , Figure 6 This is a flowchart illustrating an embodiment of optimizing the prompt statement in this application. After step S13, that is, after the server receives the operation instruction for the information processing request, it can also perform the following:

[0146] Step a1: Obtain the user account's optimization request for operation command input.

[0147] The optimization request carries optimization statements from the user account, which represent the user account's optimization intent for the operation instructions corresponding to each request intent.

[0148] For example, the optimized intent of a user's input command for "to calculate the platform revenue amount in each live broadcast session" is "not to calculate the platform revenue amount for the live broadcast session on that day".

[0149] Step a2: Based on the large language model, construct the data definition language for each optimization intent, request intent and its corresponding target data table to obtain the optimized prompt statement for each request intent.

[0150] Step a3: Based on the large language model, each optimization prompt statement is converted into a structured operation statement to obtain optimization instructions for each request intent.

[0151] Step a4: Arrange the optimized instructions sequentially based on the large language model to obtain the corresponding set of optimized instructions.

[0152] Step a5: Perform dangerous syntax verification and dangerous syntax filtering on each optimized instruction in the optimized instruction set to obtain optimized instructions for the information processing request.

[0153] The process of the large language model sequentially generating optimization prompt statements, converting them into optimization instructions, arranging the optimization instructions in order, performing dangerous syntax verification and filtering, is similar to the process in the above embodiment where the large language model sequentially generates operation prompt statements, converts them into optimization instructions, arranges the optimization instructions in order, performs dangerous syntax verification and filtering, and so on. Therefore, it will not be described in detail here.

[0154] In one exemplary embodiment, before step S14, that is, before the server executes the operation instruction, the following may also be performed:

[0155] Step 1: Obtain the index information of the target data table corresponding to the operation command.

[0156] The index information includes the table name and table type of the target data table.

[0157] Step two: Verify user account permissions based on index information and account information.

[0158] The permission verification function is used to verify whether a user account has the necessary permissions to process data in the target data table.

[0159] Step 3: If the permission verification is successful, execute the operation command.

[0160] If the permission verification fails, meaning it is confirmed that the user account does not have the functional permission to process data in the target data table, the user sends a permission request from the user terminal to the user terminal to obtain the server's permission to perform operations on the target data table.

[0161] Furthermore, in step S14, which is the process of the server executing the operation instruction after the permission verification is passed, it can specifically include the following methods:

[0162] Step 4: Obtain the user's account name and password.

[0163] Step 5: Input the account number and password into the function interface. The function interface will then sequentially perform data unlocking, data extraction, and data calculation on the information data in the target data table to obtain the task processing result for the information processing request.

[0164] In one embodiment, the data unlocking process is used by the functional interface to decrypt the information data in the target data table corresponding to the operation instruction based on the account number and password, so that the functional interface can access the information data in the target data table.

[0165] To more clearly illustrate the information processing method provided in the embodiments of this disclosure, the following describes the information processing method in detail with reference to a specific embodiment. In an exemplary embodiment, referring to... Figure 7 , Figure 7 This is a flowchart illustrating an information processing method according to another exemplary embodiment. The method is used in server 104 and specifically includes the following:

[0166] Step S21: Obtain user requirements input by the user in the terminal in the form of natural language description.

[0167] The user's request was: "I want to know the live streaming revenue of my account in the last seven days."

[0168] Step S22: Perform intent prediction on user needs to obtain the intent prediction results of input needs.

[0169] The server uses a pre-trained intent recognition model to identify the intent of the input request in order to obtain the intent inference result.

[0170] The inference result of user demand intent consists of at least one user intent, that is, at least one user intent needs to be completed in order to satisfy user demand.

[0171] Among them, the intent recognition model analyzes user needs to determine at least one user intent that needs to be fulfilled in order to meet those needs.

[0172] For example, the intent recognition model analyzes the user's statement "I want to know the live streaming revenue of my account in the last seven days" to determine the four user intents that need to be fulfilled to meet the user's needs: "determine the number of live streaming sessions in the last seven days", "calculate the amount of gift revenue in each live streaming session", "calculate the amount of platform revenue in each live streaming session", and "calculate the total amount of revenue between gift revenue and platform revenue across all live streaming sessions".

[0173] Step S23: Identify the SQL mapping type of the intent inference results and determine the SQL mapping type corresponding to each user intent.

[0174] Among them, the SQL mapping type represents the table type of the data table that the user's intent needs to be mapped to.

[0175] The table types for the data tables must include at least three types: MySQL, Hive, and FlinkSQL. In other words, the SQL mapping types must also include data tables of at least MySQL, Hive, and FlinkSQL types that correspond to user intents.

[0176] Step S24: Based on the SQL mapping type and the database types of the multiple databases to be matched, filter the multiple databases to be matched to obtain the filtered databases.

[0177] The database type is the same as the table type of the data tables stored in the database, and all data tables stored in each database are of the same type.

[0178] Step S25: Transform the intent prediction result into a first-class spatial vector, and obtain the second-class spatial vector of the table creation statement corresponding to each data table in the filtered database.

[0179] Among them, the first type of spatial vector represents the user intent in the intent prediction result, and the second type of spatial vector represents the table creation statement corresponding to the data table.

[0180] The table creation statement is used to express the data category of the data stored in the table.

[0181] The data types include integer types (INT), character types (VARCHAR), and date types (DATE).

[0182] For example, the statement to create table A1 is used to express that the data stored in table A1 is of character type, and that the character type data is the gift data of the live gifts received during the live broadcast; the statement to create table A2 is used to express that the data stored in table A2 is of time type, and that the time type data is the time data that records the user's live broadcast time; the statement to create table A3 is used to express that the data stored in table A3 is of integer type, and that the integer type data is the income data that records the user's platform revenue during the live broadcast.

[0183] Step S26: Calculate the correlation degree between the first type spatial vector corresponding to each user intent and the second type spatial vector of each data table to obtain the correlation score between each user intent and each data table.

[0184] For example, a user intent P is "to calculate the amount of gift revenue in each live stream session". The table creation statement for table A1 is used to store "gift data of live stream gifts received during the live stream", the table creation statement for table A2 is used to record "the live stream time of the user in each live stream session", and the table creation statement for table A3 is used to record "the platform revenue of the user in each live stream session". Thus, the server calculates the association score between user intent P and table A1 as 95 points, the association score between user intent P and table A2 as 10 points, and the association score between user intent P and table A3 as 65 points.

[0185] Step S27: For each user intent, select the data table with the highest pre-defined correlation score with the user intent from multiple data tables as the recommended data table, and recommend it to the user for viewing.

[0186] For example, for a user intent P, the terminal first sorts the data tables from top to bottom based on the association scores; then it selects the top two data tables from the sequence as recommended data tables; finally, it pushes the names of the recommended data tables to the user for viewing.

[0187] Step S28: Obtain the target data table selected by the user from the recommended data table.

[0188] Specifically, steps S21-S28 described above represent the process by which the server recommends a data table for generating operation tasks to the user based on the user's input requirements. Furthermore, after obtaining the recommended data table, it is necessary to generate and optimize the operation tasks based on the recommended data table.

[0189] In one exemplary embodiment, reference is made to Figure 8 , Figure 8 This is a block diagram illustrating a method for generating and optimizing operational tasks according to an exemplary embodiment. The method is used in server 104 and specifically includes the following:

[0190] Step S29: Obtain user requirements input in the form of natural language description in the terminal.

[0191] Among them, user requirements include either the first type of requirement, which is to generate work tasks based on the target data table, or the second type of requirement, which is to optimize the work tasks.

[0192] Step S30: If the user requirement is the first type of user requirement, determine whether the user selects the target data table from the filtered database.

[0193] If the user does not select a target data table, proceed to step 30; if the user has already selected a target data table, proceed to step 31.

[0194] Step S31: For each user intent, based on the association scores of each data table in the filtered database, select the data table with the highest association score as the target data table.

[0195] Step S32: Obtain the data definition language of the target data table based on the table information index of the target data table.

[0196] The table information of the target data table includes the table name, table number, etc.

[0197] The data definition language of a data table is used to characterize the structure of the data table, the meaning of its fields, and the relationships between the fields.

[0198] The server indexes the data definition language (DDL) of the target data table from the metadata system based on the table information of the target data table.

[0199] The metadata system is used to store the attribute data of the data tables, which includes the data definition language of the data tables.

[0200] Step S33: Based on the data definition language of each user intent and its corresponding target data table, construct a task prompt for each user intent.

[0201] Among them, the task prompt is the task node to be performed on the target data table according to the user's intention, and each task prompt is expressed in natural language.

[0202] For example, the task prompt for the user intent "determine the live stream sessions in the last seven days" is "extract the user account's live stream records from XX month XX day to XX month XX day from the target data table M1, and then determine the live stream sessions corresponding to each live stream record"; the task prompt for the user intent "calculate the gift revenue in each live stream session" is "extract the user account's gift revenue received in each live stream session from XX month XX day to XX month XX day from the target data table M2, and then calculate the total gift revenue between the gift revenue amounts of each live stream session"; the task prompt for the user intent "calculate the platform revenue in each live stream session" is "extract the user account's platform revenue earned in each live stream session from XX month XX day to XX month XX day from the target data table M3, and then calculate the total platform revenue between the platform revenue amounts of each live stream session"; the task prompt for the user intent "calculate the total revenue between gift revenue and platform revenue across all live stream sessions" is "calculate the total revenue between the total platform revenue and total gift revenue of the user account in each live stream session from XX month XX day to XX month XX day".

[0203] Step S34: Input the task prompts of each user intent into the large model for SQL transformation and order integration to obtain the output SQL result.

[0204] Among them, SQL transformation is used to convert task prompts into corresponding SQL operation tasks, and order integration is used to arrange all SQL operation tasks in sequence to obtain the output SQL results.

[0205] The SQL operation task is a structured executable program.

[0206] For example, the large model first converts the user intent "to count the number of live streams in the last seven days" task prompt N1 into SQL operation task Q1, the user intent "to count the amount of gift revenue in each live stream" task prompt N2 into SQL operation task Q2, the user intent "to count the amount of platform revenue in each live stream" task prompt N3 into SQL operation task Q3, and the user intent "to count the total amount of revenue between gift revenue and platform revenue across all live streams" task prompt N4 into SQL operation task Q4; then, the SQL operation tasks Q1, Q2, Q3, and Q4 are arranged in sequence.

[0207] Step S35: Perform dangerous syntax validation and filtering on the SQL results to obtain the output SQL results.

[0208] Each SQL operation task includes various types of execution functions (including command execution functions, database operation functions, file operation functions, etc.) and operation characters. After generating each SQL operation task, it is necessary to protect against SQL injection vulnerabilities (such as SQL injection statements like drop table) to ensure the legitimacy and security of subsequent operations on the data table.

[0209] The process of validating and filtering dangerous syntax in SQL results includes setting corresponding security filtering functions for the command execution functions in each SQL operation task. These security filtering functions can alert users to and automatically modify the data content in SQL operation tasks that exhibit validation anomalies (e.g., disallowing DDL operations on MySQL and Hive tables, while allowing DDL operations on FlinkSQL tables), ensuring that the request parameters comply with secure execution standards.

[0210] Step S36: If the user requirement is the second type of user requirement, obtain the SQL result and the optimization requirements for the SQL result.

[0211] The optimization requirements can include user-inputted optimization requirements for SQL results, or system-default optimization requirements for SQL results.

[0212] Step S37: Based on the optimization requirements, determine whether there are any SQL operation tasks to be optimized in the SQL results.

[0213] If no SQL operation task to be optimized is found in the SQL results, a reminder message will be sent to the user to remind them to upload the SQL operation task to be optimized.

[0214] Step S38: If there is an SQL operation task to be optimized in the SQL result, obtain the data definition language of the target data table based on the table information index of the target data table corresponding to the SQL operation task to be optimized.

[0215] Step S39: Based on the optimization requirements, the user intent corresponding to the user's input requirements, and the data definition language of the target data table, construct optimization suggestions for the SQL operation task to be optimized.

[0216] Among them, the optimization prompt is the optimized task prompt reconstructed for the SQL operation task to be optimized.

[0217] Step S40: Input the SQL operation task to be optimized into the large model for SQL transformation and sequence integration to obtain the optimized SQL output.

[0218] Step S41: Perform dangerous syntax checks and filtering on the optimized SQL results to obtain the output executable SQL results.

[0219] Steps S29-S41 described above constitute the process of the server generating and optimizing the operation tasks. Furthermore, after obtaining the optimized operation tasks, they still need to be executed.

[0220] In one exemplary embodiment, reference is made to Figure 9 , Figure 9 This is a block diagram illustrating a method for executing an operation task according to an exemplary embodiment. The method is used in server 104 and specifically includes the following:

[0221] Step S42: Obtain the executable SQL result and the user ID of the executing user.

[0222] Step S43: Perform dangerous syntax checks and filtering on the executable SQL results to obtain normal executable SQL results.

[0223] If there are abnormal SQL operation tasks in the executable SQL results, the data content contained in the abnormal SQL operation tasks will be alerted and automatically modified to make the SQL operation tasks comply with the safe execution standards.

[0224] Step S44: Based on the user ID, determine whether the executing user has the permission to operate on the target data table corresponding to each SQL operation task in the executable SQL result.

[0225] Specifically, if the user does not have permission to operate on the target data table, the user submits an operation permission request to the server in order to obtain the server to grant the user the permission to operate on the target data table.

[0226] Step S45: If the executing user has the necessary permissions to operate on the target data table, retrieve the executing user's account and password from the database, and determine the operation type of each SQL operation task in the executable SQL results.

[0227] The SQL operation tasks include operation types such as query type (SELECT), insert type (INSERT), update type (UPDATE), or delete type (DELETE).

[0228] Step S46: Based on the execution account and password, decrypt the target data table corresponding to each SQL operation task.

[0229] Step S47: Based on the operation type of each SQL operation task, call the corresponding interface type function interface to extract and process data from the decrypted data table, and obtain the SQL result for the executable SQL result output.

[0230] The interface type of the functional interface matches the operation type of the SQL operation task.

[0231] The functional interface is used to connect to various execution engines or platform systems to output or process the stored data in the data table offline or in real time.

[0232] Step S48: Convert the SQL results output for the executable SQL results into a visual report and send it to the user terminal for display to the executing user.

[0233] In this way, on the one hand, this solution determines the target data table that matches the user's request intent through the user's dialogue statements. Then, using the request intent and the target data table, it obtains and executes operation instructions for the information processing request, thereby optimizing the information processing flow, effectively improving the efficiency of information processing based on the user's request intent, and reducing the consumption of manpower and resources. On the other hand, unlike existing technologies, this solution uses a pre-trained language model to output a structured operation language that matches the attribute information of the user's request intent and the target data table. Data processing is then performed according to this structured operation language to obtain the processing result, thus optimizing the generation method of operation instructions, effectively improving the rationality of the generated operation instructions, and reducing the deviation between the final processing result and the user's actual request intent, resulting in a better user experience.

[0234] It should be understood that, although Figures 2-9The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figures 2-9 At least some of the steps in the process may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least some of the steps or stages in other steps.

[0235] It is understood that the same / similar parts between the various embodiments of the methods described above in this specification can be referred to each other. Each embodiment focuses on the differences from other embodiments, and relevant parts can be referred to the description of other method embodiments.

[0236] Figure 10 This is a block diagram of an information processing device provided in an embodiment of this application. (Refer to...) Figure 10 The information processing device 10 includes: a request acquisition unit 11, a target data table unit 12, an instruction output unit 13, and an instruction execution unit 14.

[0237] The request acquisition unit 11 is configured to execute an information processing request triggered by acquiring a user account; the information processing request carries a dialogue statement of the user account, and the dialogue statement represents the request intent of the information processing request.

[0238] The target data table unit 12 is configured to perform a process based on the dialogue statement to select a target data table from multiple data tables stored in multiple databases; the information data stored in the target data table matches the request intent.

[0239] The instruction output unit 13 is configured to input the request intent and the attribute information of the target data table into a pre-trained large language model to output an operation instruction for the information processing request; the operation instruction is based on a structured operation language representation.

[0240] The instruction execution unit 14 is configured to execute the operation instruction to call the corresponding function interface to process the information data in the target data table and obtain the task processing result for the information processing request.

[0241] In some embodiments, the step of filtering a target data table from multiple data tables stored in multiple databases based on the dialogue statement includes:

[0242] The intent of the dialogue statement is identified to obtain the request intent for the dialogue statement;

[0243] Based on the request type of the request intent, candidate databases of the corresponding database type are selected from multiple databases; the request type of the request intent matches the database type of the candidate database.

[0244] The request intent is correlated and matched with the candidate data tables in the candidate database to determine the target data table that matches the request intent from the candidate database.

[0245] In some embodiments, the step of associating and matching the request intent with candidate data tables in the candidate database to determine a target data table that matches the request intent from the candidate database includes:

[0246] The request intent is converted into a first type of spatial vector, and the table creation statements of each candidate data table in the candidate database are converted into a second type of spatial vector; the table creation statements are used to express the data type of the information data stored in the candidate data tables.

[0247] The correlation degree is calculated between the first type of spatial vector and each of the second type of spatial vectors to obtain the correlation score for each of the candidate data tables;

[0248] Among the candidate data tables, the candidate data table with the highest correlation score is selected as the target data table.

[0249] In some embodiments, the number of request intents for the dialogue statement includes at least two;

[0250] The step of inputting the request intent and the attribute information of the target data table into a pre-trained large language model to output operation instructions for the information processing request includes:

[0251] Based on the attribute information of the target data table corresponding to each of the aforementioned request intentions, a data definition language is determined for each of the aforementioned target data tables; the data definition language is used to characterize the structure, field meanings, and relationships between fields of the target data table.

[0252] Based on the large language model, instructions are generated from the data definition language of each request intent and its corresponding target data table to obtain operation instructions for the information processing request.

[0253] In some embodiments, the step of generating instructions based on the data definition language of each request intent and its corresponding target data table according to the large language model to obtain operation instructions for the information processing request includes:

[0254] Based on the large language model, prompts are constructed for each of the request intentions and their corresponding target data tables to obtain operation prompt statements for each of the request intentions.

[0255] Based on the large language model, each of the operation prompt statements is converted into a structured operation statement to obtain operation instructions for each of the request intentions;

[0256] Based on the large language model, the operation instructions are arranged sequentially to obtain a corresponding set of operation instructions, and the set of operation instructions is used as the operation instructions for the information processing request.

[0257] In some embodiments, the information processing request also carries account information of the user account;

[0258] Before executing the operation instruction, the method further includes:

[0259] Obtain the index information of the target data table corresponding to the operation instruction;

[0260] The user account is subject to permission verification based on the index information and the account information; the permission verification is used to verify whether the user account has the functional permission to process data in the target data table.

[0261] If the permission verification passes, the operation instruction is executed.

[0262] In some embodiments, executing the operation instructions includes:

[0263] Identify the operation type of the operation instruction and the table type of the target data table; the table type includes offline type and real-time type;

[0264] When the table type is offline, the function interface corresponding to the operation type is called to perform offline data processing on the information data in the target data table.

[0265] When the table type is real-time, the function interface corresponding to the operation type is called to perform real-time data processing on the information data in the target data table.

[0266] In some embodiments, the step of calling the corresponding function interface to process the information data in the target data table and obtaining the task processing result for the information processing request includes:

[0267] If the permission verification passes, obtain the user's account number and password;

[0268] The account number and password are input into the function interface, so that the information data in the target data table can be processed sequentially through the function interface to perform data unlocking, data extraction and data calculation, so as to obtain the task processing result for the information processing request.

[0269] In some embodiments, prior to executing the operation instruction, the method further includes:

[0270] The operation instructions are subjected to dangerous syntax verification and dangerous syntax filtering to obtain the processed operation instructions;

[0271] The dangerous syntax verification process is used to verify whether there are abnormal functions and / or abnormal characters in the operation instruction;

[0272] The dangerous syntax filtering process is used to modify abnormal functions in the operation instructions into normal functions and / or remove abnormal characters in the operation instructions.

[0273] In some embodiments, after receiving the operation instruction for the information processing request, the method further includes:

[0274] Obtain the user account's optimization request for the operation instruction input; the optimization request carries the user account's optimization statement, the optimization statement representing the user account's optimization intention for each operation instruction corresponding to the request intention;

[0275] Based on the large language model, the data definition language of each optimization intent, request intent and its corresponding target data table is constructed to obtain optimized prompt statements for each request intent;

[0276] Based on the large language model, each of the optimized prompt statements is converted into a structured operation statement to obtain optimized instructions for each of the request intentions;

[0277] Based on the large language model, the optimized instructions are arranged sequentially to obtain the corresponding set of optimized instructions;

[0278] Each optimization instruction in the optimization instruction set is subjected to dangerous syntax verification and dangerous syntax filtering to obtain an optimization instruction for the information processing request.

[0279] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0280] In one embodiment, a computer device is provided, which, depending on the program logic to be applied, can be an electronic device or a server, and its internal structure diagram can be as follows: Figure 11 As shown. The computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus.

[0281] The processor of this computer device provides computing and control capabilities. The memory of the computer device includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed, it implements an information processing method. The display screen of the computer device can be an LCD screen or an e-ink display screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad located on the casing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0282] Those skilled in the art will understand that Figure 11 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer equipment (including servers and electronic devices) to which the present application is applied. Specific computer equipment may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0283] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed:

[0284] Obtain information processing requests triggered by user accounts; the information processing requests carry dialogue statements of the user accounts, and the dialogue statements represent the request intent of the information processing requests.

[0285] Based on the dialogue statement, a target data table is selected from multiple data tables stored in multiple databases; the information data stored in the target data table matches the request intent.

[0286] The request intent and the attribute information of the target data table are input into a pre-trained large language model to output operation instructions for the information processing request; the operation instructions are based on structured operation language representations.

[0287] The operation instruction is executed to call the corresponding function interface to process the information data in the target data table and obtain the task processing result for the information processing request.

[0288] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0289] Obtain information processing requests triggered by user accounts; the information processing requests carry dialogue statements of the user accounts, and the dialogue statements represent the request intent of the information processing requests.

[0290] Based on the dialogue statement, a target data table is selected from multiple data tables stored in multiple databases; the information data stored in the target data table matches the request intent.

[0291] The request intent and the attribute information of the target data table are input into a pre-trained large language model to output operation instructions for the information processing request; the operation instructions are based on structured operation language representations.

[0292] The operation instruction is executed to call the corresponding function interface to process the information data in the target data table and obtain the task processing result for the information processing request.

[0293] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0294] Those skilled in the art will understand that embodiments of this application can provide information processing methods, information processing apparatus, servers, computer equipment, computer-readable storage media, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer program instructions (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0295] This application is described with reference to flowchart illustrations and / or block diagrams of information processing methods, information processing apparatuses, servers, computer devices, computer-readable storage media, or computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by the computer program product. These computer program products can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that program instructions executable by the processor of the computer or other programmable data processing apparatus generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0296] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile 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. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0297] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. An information processing method, characterized in that, The method includes: Obtain information processing requests triggered by user accounts; the information processing requests carry dialogue statements of the user accounts, and the dialogue statements represent the request intent of the information processing requests; the number of request intents for the dialogue statements includes at least two. Based on the dialogue statement, a target data table is selected from multiple data tables stored in multiple databases; the information data stored in the target data table matches the request intent. The request intent and the attribute information of the target data table are input into a pre-trained large language model to output operation instructions for the information processing request. The operation instructions are based on a structured operation language representation, including: determining the data definition language of each target data table based on the attribute information of the target data table corresponding to each request intent; the data definition language is used to represent the structure, field meaning, and inter-field relationships of the target data table; constructing prompts for each request intent and its corresponding target data table based on the large language model to obtain operation prompt statements for each request intent; converting each operation prompt statement into a structured operation statement based on the large language model to obtain operation instructions for each request intent; arranging each operation instruction in sequence based on the large language model to obtain a corresponding set of operation instructions, and using the set of operation instructions as the operation instructions for the information processing request. The operation instruction is executed to call the corresponding function interface to process the information data in the target data table and obtain the task processing result for the information processing request.

2. The method according to claim 1, characterized in that, The step of filtering the target data table from multiple data tables stored in multiple databases based on the dialogue statement includes: The intent of the dialogue statement is identified to obtain the request intent for the dialogue statement; Based on the request type of the request intent, candidate databases of the corresponding database type are selected from multiple databases; the request type of the request intent matches the database type of the candidate database. The request intent is correlated and matched with the candidate data tables in the candidate database to determine the target data table that matches the request intent from the candidate database.

3. The method according to claim 2, characterized in that, The step of associating and matching the request intent with candidate data tables in the candidate database to determine the target data table that matches the request intent from the candidate database includes: The request intent is converted into a first type of spatial vector, and the table creation statements of each candidate data table in the candidate database are converted into a second type of spatial vector; the table creation statements are used to express the data type of the information data stored in the candidate data tables. The correlation degree is calculated between the first type of spatial vector and each of the second type of spatial vectors to obtain the correlation score for each of the candidate data tables; Among the candidate data tables, the candidate data table with the highest correlation score is selected as the target data table.

4. The method according to any one of claims 1-3, characterized in that, The information processing request also carries the user account information; Before executing the operation instruction, the method further includes: Obtain the index information of the target data table corresponding to the operation instruction; The user account is subject to permission verification based on the index information and the account information; the permission verification is used to verify whether the user account has the functional permission to process data in the target data table. If the permission verification passes, the operation instruction is executed.

5. The method according to claim 4, characterized in that, The execution of the operation instruction includes: Identify the operation type of the operation instruction and the table type of the target data table; the table type includes offline type and real-time type; When the table type is offline, the function interface corresponding to the operation type is called to perform offline data processing on the information data in the target data table. When the table type is real-time, the function interface corresponding to the operation type is called to perform real-time data processing on the information data in the target data table.

6. The method according to claim 4, characterized in that, The process of calling the corresponding function interface to process the information data in the target data table and obtaining the task processing result for the information processing request includes: If the permission verification passes, obtain the user's account number and password; The account number and password are input into the function interface, so that the information data in the target data table can be processed sequentially through the function interface to perform data unlocking, data extraction and data calculation, so as to obtain the task processing result for the information processing request.

7. The method according to claim 1, characterized in that, Before executing the operation instruction, the method further includes: The operation instructions are subjected to dangerous syntax verification and dangerous syntax filtering to obtain the processed operation instructions; The dangerous syntax verification process is used to verify whether there are abnormal functions and / or abnormal characters in the operation instruction; The dangerous syntax filtering process is used to modify abnormal functions in the operation instructions into normal functions and / or remove abnormal characters in the operation instructions.

8. The method according to claim 1, characterized in that, After receiving the operation instruction for the information processing request, the process also includes: Obtain the user account's optimization request for the operation instruction input; the optimization request carries the user account's optimization statement, the optimization statement representing the user account's optimization intention for each operation instruction corresponding to the request intention; Based on the large language model, the data definition language of each optimization intent, request intent and its corresponding target data table is constructed to obtain optimized prompt statements for each request intent; Based on the large language model, each of the optimized prompt statements is converted into a structured operation statement to obtain optimized instructions for each of the request intentions; Based on the large language model, the optimized instructions are arranged sequentially to obtain the corresponding set of optimized instructions; Each optimization instruction in the optimization instruction set is subjected to dangerous syntax verification and dangerous syntax filtering to obtain an optimization instruction for the information processing request.

9. An information processing device, characterized in that, include: The request retrieval unit is configured to execute an information processing request triggered by retrieving a user account; The information processing request carries a dialogue statement of the user account, and the dialogue statement represents the request intent of the information processing request. The number of request intents for the aforementioned dialogue statement includes at least two; The target data table unit is configured to perform a selection of a target data table from multiple data tables stored in multiple databases based on the dialog statement. The information stored in the target data table matches the request intent; The instruction output unit is configured to input the request intent and the attribute information of the target data table into a pre-trained large language model to output an operation instruction for the information processing request. The operation instructions are based on a structured operation language representation; including: determining the data definition language of each target data table based on the attribute information of the target data table corresponding to each request intent; the data definition language is used to represent the structure, field meanings, and relationships between fields of the target data table; constructing prompts for each request intent and its corresponding target data table based on the large language model to obtain operation prompt statements for each request intent; converting each operation prompt statement into a structured operation statement based on the large language model to obtain operation instructions for each request intent; arranging each operation instruction in sequence based on the large language model to obtain a corresponding set of operation instructions, and using the set of operation instructions as the operation instructions for the information processing request; The instruction execution unit is configured to execute the operation instruction to call the corresponding function interface to process the information data in the target data table and obtain the task processing result for the information processing request.

10. A computer device, characterized in that, include: processor; Memory for storing the executable instructions of the processor; The processor is configured to execute the executable instructions to implement the information processing method as described in any one of claims 1 to 8.

11. A computer-readable storage medium comprising program data, characterized in that, When the program data is executed by the processor of a computer device, the computer device is able to perform the information processing method as described in any one of claims 1 to 8.

Images