Task processing method, device and system

Abstract

The invention provides a task processing method, device and system, and relates to the technical field of data processing, in particular to the technical fields of artificial intelligence, natural language processing and the like. The method comprises the following steps: calling a preset large language model to decompose a task instruction to obtain at least one target processing instruction; in response to a first processing instruction belonging to the reference processing instruction in the target processing instruction, retrieving the first processing instruction based on the association relationship, and determining a target processing tool corresponding to the first processing instruction; in response to a second processing instruction which does not belong to the reference processing instruction in the target processing instruction, constructing a target processing tool corresponding to the second processing instruction; and calling a target processing tool corresponding to the target processing instruction, and executing the target processing instruction to obtain a target processing result corresponding to the task instruction. The task processing efficiency and accuracy can be improved, the cross-task generalization ability is improved, and the method and device adapt to changes of new tasks and new business environments.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to technical fields such as natural language processing and artificial intelligence. Background Art

[0002] In the related technology, the task processing system mainly involves processing known and specific tasks. During the task processing process, it relies on traditional reinforcement learning algorithms, which are implemented through a large number of specific task training and evaluation. However, for dynamically changing business environments, unknown or new task instructions may appear.

[0003] Therefore, how to improve the efficiency and accuracy of task processing, improve cross-task generalization capabilities, and adapt to changes in new tasks and new business environments has become one of the important research directions. Summary of the invention

[0004] The present disclosure provides a task processing method, device and system.

[0005] According to one aspect of the present disclosure, there is provided a task processing method, comprising:

[0006] Receive a task instruction, and call a preset large language model to decompose the task instruction to obtain at least one target processing instruction;

[0007] Obtaining the association relationship between the reference processing instruction and the reference processing tool in the preset database;

[0008] In response to the existence of a first processing instruction belonging to the reference processing instruction in the target processing instructions, the first processing instruction is retrieved based on the association relationship to determine a target processing tool corresponding to the first processing instruction;

[0009] In response to the presence of a second processing instruction in the target processing instruction that does not belong to the reference processing instruction, constructing a target processing tool corresponding to the second processing instruction;

[0010] The target processing tool corresponding to the target processing instruction is called, and the target processing instruction is executed to obtain the target processing result corresponding to the task instruction.

[0011] According to another aspect of the present disclosure, there is provided a task processing device, comprising:

[0012] A first acquisition module is used to receive a task instruction and call a preset large language model to decompose the task instruction to obtain at least one target processing instruction;

[0013] A second acquisition module is used to acquire the association relationship between the reference processing instruction and the reference processing tool in the preset database;

[0014] A determination module, configured to retrieve the first processing instruction based on the association relationship in response to the existence of a first processing instruction belonging to the reference processing instruction in the target processing instruction, and determine a target processing tool corresponding to the first processing instruction;

[0015] A first processing module, configured to construct a target processing tool corresponding to the second processing instruction in response to the presence of a second processing instruction in the target processing instruction that does not belong to the reference processing instruction;

[0016] The second processing module is used to call the target processing tool corresponding to the target processing instruction, execute the target processing instruction, and obtain the target processing result corresponding to the task instruction.

[0017] According to another aspect of the present disclosure, there is provided an electronic device, including at least one processor, and

[0018] a memory communicatively connected to at least one processor; wherein,

[0019] The memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor so that the at least one processor can execute the task processing method of the embodiment of the first aspect of the present disclosure.

[0020] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable a computer to execute the task processing method according to the embodiment of the first aspect of the present disclosure.

[0021] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the computer program implements the steps of the task processing method of the embodiment of the first aspect of the present disclosure.

[0022] According to another aspect of the present disclosure, a task processing system is provided, including a task processing device according to the embodiment of the second aspect of the present disclosure as described above.

[0023] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0025] Figure 1 is a flowchart of a task processing method according to an embodiment of the present disclosure;

[0026] Figure 2 is a flowchart of a task processing method according to an embodiment of the present disclosure;

[0027] Figure 3 is a flowchart of a task processing method according to an embodiment of the present disclosure;

[0028] Figure 4 is a flowchart of a task processing method according to an embodiment of the present disclosure;

[0029] Figure 5 is a flowchart of a task processing method according to an embodiment of the present disclosure;

[0030] Figure 6 is a structural diagram of a task processing device according to an embodiment of the present disclosure;

[0031] Figure 7 The block diagram is a block diagram of an electronic device for implementing the method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0032] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0033] The embodiments of the present disclosure relate to technical fields such as data processing and artificial intelligence.

[0034] Data processing: Data is a form of expression of facts, concepts or instructions, which can be processed by manual or automated devices. After data is interpreted and given a certain meaning, it becomes information. Data processing is the collection, storage, retrieval, processing, transformation and transmission of data. The basic purpose of data processing is to extract and derive valuable and meaningful data for certain specific people from a large amount of data that may be chaotic and difficult to understand. Data processing is a basic link in system engineering and automatic control. Data processing runs through all areas of social production and social life. The development of data processing technology and the breadth and depth of its application have greatly affected the process of human social development.

[0035] Artificial Intelligence (AI) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence.

[0036] Natural Language Processing (NLP) is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that can achieve effective communication between people and computers using natural language. Natural language processing is a science that integrates linguistics, computer science, and mathematics. Therefore, research in this field will involve natural language, that is, the language people use in daily life, so it is closely related to the study of linguistics, but there are important differences. Natural language processing is not a general study of natural language, but the development of computer systems that can effectively realize natural language communication, especially the software systems therein.

[0037] The following are explanations of the terms used in this disclosure:

[0038] Large Language Model (LLM) refers to a deep learning model trained with a large amount of text data, which can generate natural language text or understand the meaning of language text. Large language models can handle a variety of natural language tasks, such as text classification, question answering, and dialogue, and are an important path to artificial intelligence.

[0039] Meta-tools are tools used for meta-learning. Meta-Learning, also known as learning how to learn, is a machine learning method that aims to enable the model to better adapt to new tasks through the experience of learning tasks.

[0040] The task processing method, device and system disclosed in the present invention are described below with reference to the accompanying drawings.

[0041] Figure 1 is a flowchart of a task processing method according to an embodiment of the present disclosure. Figure 1 As shown, the method comprises the following steps:

[0042] S101, receiving a task instruction, calling a preset large language model to decompose the task instruction, and obtaining at least one target processing instruction.

[0043] For example, a task instruction of a terminal device or a preset application program may be received, wherein the task instruction includes one or more processing instructions. In the embodiment of the present disclosure, a preset large language model LLM is called to decompose the task instruction to obtain one or more target processing instructions. For example, a large language model may be called to decompose the task instruction to obtain one or more subtask instructions, and then each subtask instruction is decomposed to obtain one or more grandchild task instructions. Similarly, the decomposition is continued according to the above steps until an instruction of a preset task granularity is obtained, which is the target processing instruction.

[0044] For example, if the task instruction is "I want to go to Beijing", the preset large language model is called to decompose the task instruction and obtain five target processing instructions, including "determine the departure point", "determine the arrival time", "query transportation", "select transportation", and "purchase tickets".

[0045] S102, obtaining the association relationship between the reference processing instructions and the reference processing tools in the preset database.

[0046] In some implementations, the association relationship indicates a reference processing tool associated with the reference processing instruction; in some implementations, the association relationship indicates that the reference processing instruction is associated with multiple reference processing tools, which means that the execution of the reference processing instruction requires calling multiple reference processing tools in sequence.

[0047] In some embodiments, the processing tool may be a preset algorithm, in some embodiments, the processing tool may be a pre-deployed model, or a program interface, in other embodiments, the processing tool may also be other tools, which are not limited in the embodiments of the present disclosure. With the help of the tool, the system can use external resources (such as databases, network applications, etc.) for reasoning and planning, thereby expanding its action space. For example, external data can be called through the processing tool. This process provides a richer and more reliable source of information for the task processing system, which promotes the improvement of the quality of decision-making and action.

[0048] In some embodiments, all target processing instructions belong to reference processing instructions. In some embodiments, all target processing instructions do not belong to reference processing instructions. In some embodiments, some target processing instructions belong to reference processing instructions and some do not belong to reference processing instructions. The embodiments of the present disclosure are not limited to this.

[0049] Optionally, the reference processing tool is stored in a pre-deployed tool library, and the reference processing tool can be called from the preset tool library later.

[0050] S103, in response to the existence of a first processing instruction belonging to the reference processing instruction in the target processing instructions, searching the first processing instruction based on the association relationship, and determining a target processing tool corresponding to the first processing instruction.

[0051] In some embodiments, the target processing instruction belongs in whole or in part to the reference processing instruction. For the first processing instruction in the target processing instruction that belongs to the reference processing instruction, each first processing instruction is retrieved based on the association relationship, and the reference processing tool associated with the first processing instruction is determined as the target processing tool corresponding to the first processing instruction.

[0052] Optionally, taking the first processing instruction as “buy a ticket” as an example, the target processing tool corresponding to the first processing instruction is a purchase and payment interface.

[0053] S104 , in response to the existence of a second processing instruction that does not belong to the reference processing instruction in the target processing instruction, construct a target processing tool corresponding to the second processing instruction.

[0054] In some embodiments, the target processing instruction is not entirely or partially a reference processing instruction, and a processing tool can be constructed through the generation capability of LLM, and the constructed processing tool is more modular, and the input and output formats are more suitable for system requirements. LLM can generate an executable program or integrate existing tools based on the second processing instruction to obtain the target processing tool, thereby improving the efficiency and flexibility of task processing.

[0055] Successfully created tools can be stored in the skill library learning and management module to promote tool sharing and reuse.

[0056] S105, calling a target processing tool corresponding to the target processing instruction, executing the target processing instruction, to obtain a target processing result corresponding to the task instruction.

[0057] In some implementations, there is one target processing instruction, and the target processing tool corresponding to the target processing instruction is directly called to execute the target processing instruction to obtain the target processing result of the target processing instruction. Optionally, the target processing result can also be text-processed based on the large language model, such as language conversion or summarization, to update the target processing result.

[0058] In some embodiments, there are two or more target processing instructions, and it is necessary to obtain the decomposition order of each target processing instruction in step S101, call the target processing tool corresponding to each target processing instruction based on the decomposition order, execute the target processing instructions in sequence, and generate candidate processing results for each target processing instruction. Optionally, a language model can be called to integrate the candidate processing results of each target processing instruction based on the decomposition order to obtain the target processing result.

[0059] In the embodiment of the present disclosure, a preset large language model is called to decompose the task instruction to obtain at least one target processing instruction, the association relationship between the reference processing instruction and the reference processing tool in the preset database is obtained, the first processing instruction is retrieved based on the association relationship, and the target processing tool corresponding to the first processing instruction is determined, which can improve the efficiency of task processing, and the target processing tool corresponding to the second processing instruction is constructed, which can improve the accuracy of the newly appeared task processing instruction; the target processing tool corresponding to the target processing instruction is called, and the target processing instruction is executed to obtain the target processing result corresponding to the task instruction. The present disclosure can flexibly organize and coordinate the use of multiple tools, improve the ability of cross-task generalization, adapt to the changes of new tasks and new business environments, and complete complex tasks.

[0060] Figure 2is a flowchart of a task processing method according to an embodiment of the present disclosure. Figure 2 As shown, for a second processing instruction in the target processing instruction that does not belong to the reference processing instruction, before constructing a target processing tool corresponding to the second processing instruction, the following steps are also included:

[0061] S201, obtaining attribute information of a reference processing tool, the attribute information including tool name, application scope and calling method.

[0062] In some implementations, attribute information of each reference processing tool is obtained. Taking the reference processing tool as a payment interface as an example, the tool name may be "ticket fare payment interface", the application scope may be "domestic payment", and the calling method may be "interface call".

[0063] S202, traverse the second processing instructions, and for the third processing instruction currently traversed, match the third processing instruction based on the attribute information of the reference processing tool to determine whether there is a target processing tool corresponding to the third processing instruction in the reference processing tool.

[0064] In some embodiments, intent recognition is performed on the third processing instruction to determine the first intent of the third processing instruction, intent recognition is performed on each reference processing tool based on the application scope and calling method of each reference processing tool to determine the second intent of each reference processing tool, the first intent is matched with each second intent to obtain a first matching degree for each reference processing tool, the third processing instruction is matched with the tool name of each reference processing tool to obtain a second matching degree for each reference processing tool, and based on the first matching degree and the second matching degree of each reference processing tool and a preset priority of each reference processing tool, it is determined whether there is a target processing tool corresponding to the third processing instruction in the reference processing tool.

[0065] The matching process is the result of comprehensive judgment of multiple factors. For example, the first matching degree and the second matching degree can be weighted to obtain the target matching degree. The reference processing tool with a target matching degree greater than the preset first matching degree threshold and the highest priority is used as the target processing tool corresponding to the third processing instruction. Otherwise, it is determined that the target processing tool corresponding to the third processing instruction does not exist in the reference processing tools.

[0066] For example, a reference processing tool whose first matching degree is greater than a preset second matching degree threshold, whose second matching degree is greater than a preset third matching degree threshold and whose priority is the highest can be used as the target processing tool corresponding to the third processing instruction; otherwise, it is determined that there is no target processing tool corresponding to the third processing instruction in the reference processing tools.

[0067] S203 , in response to the existence of a target processing tool corresponding to the third processing instruction in the reference processing tool, establishing an association relationship between the third processing instruction and the target processing tool corresponding to the third processing instruction.

[0068] Optionally, for the third processing instruction matched to the target processing tool, an association relationship may be established between the third processing instruction and the target processing tool corresponding to the third processing instruction, so as to quickly obtain the processing tool corresponding to the processing instruction based on the association relationship next time.

[0069] S204, clear the third processing instruction from the second processing instruction.

[0070] It should be noted that if the target processing tool corresponding to the third processing instruction can be matched from the reference processing tool, there is no need to construct the target processing tool corresponding to the third processing instruction. Therefore, the third processing instruction is cleared from the second processing instruction to update the second processing instruction.

[0071] In the disclosed embodiment, for the second processing instruction that does not belong to the reference processing instruction, the database is queried, and if there is a related processing tool that can be directly executed, it is directly called for execution, which can adapt to changes in new tasks and new business environments and improve the efficiency of task processing.

[0072] In some implementations, in order to enrich the processing tools and improve their adaptability and generalization, the processing tool usage skills learned in a specific environment can be transferred to more general scenarios to obtain reference processing tools with different application ranges. For example, scenario transfer can be achieved through meta-tool learning.

[0073] Figure 3 is a flowchart of a task processing method according to an embodiment of the present disclosure. Figure 1 As shown, the method comprises the following steps:

[0074] S301, receiving a task instruction, calling a preset large language model to decompose the task instruction, and obtaining at least one target processing instruction.

[0075] S302, obtaining the association relationship between the reference processing instruction and the reference processing tool in the preset database.

[0076] S303: In response to the existence of a first processing instruction belonging to the reference processing instruction in the target processing instructions, the first processing instruction is retrieved based on the association relationship to determine a target processing tool corresponding to the first processing instruction.

[0077] For the introduction of step S301 to step S302, reference may be made to the relevant contents in the above embodiment, which will not be repeated here.

[0078] S304, in response to the existence of a second processing instruction that does not belong to the reference processing instruction in the target processing instruction, calling a large language model to analyze the second processing instruction and a preset prompt word to generate a candidate code.

[0079] LLM itself has the ability to generate code, so in some implementations, a large language model can be called to analyze the second processing instruction and the preset prompt word to generate candidate code. Optionally, meta-tool learning can be introduced to enable LLM to autonomously design and generate new candidate code.

[0080] S305, pre-setting a deployment tool under a preset business environment to package and deploy the candidate codes.

[0081] After the candidate code is generated, a deployment tool can be preset in the preset business environment. The deployment tool packages the input candidate code in the form of an interface and deploys it in the business environment.

[0082] S306, calling a debugging interface to debug the candidate code, and obtaining a target processing tool corresponding to the second processing instruction.

[0083] The debugging interface is called to view the returned result for debugging. Debugging means modifying the program according to the returned result. The modification can be performed by LLM. The deployed program is the target processing tool corresponding to the second processing instruction.

[0084] It should be noted that in the embodiment of the present application, new processing tools are independently designed and generated, the content of the tool library is continuously enriched, the target processing tools corresponding to the second processing instructions are stored, and an association relationship is established between the second processing instructions and the target processing tools corresponding to the second processing instructions. This can facilitate the improvement of task processing efficiency the next time the task is processed.

[0085] S307, calling the target processing tool corresponding to the target processing instruction, executing the target processing instruction, to obtain the target processing result corresponding to the task instruction.

[0086] For the introduction of step S307, please refer to the relevant content in the above embodiment, which will not be repeated here.

[0087] In the disclosed embodiment, in response to the presence of a second processing instruction in the target processing instruction that does not belong to the reference processing instruction, a large language model is called to analyze the second processing instruction and a preset prompt word to generate candidate codes, a deployment tool is preset under a preset business environment, the candidate codes are packaged and deployed, and a debugging interface is called to debug the candidate codes to obtain a target processing tool corresponding to the second processing instruction. This can adapt to changes in new tasks and new business environments, improve the accuracy of newly emerging task processing instructions, and be able to flexibly respond to unknown and complex tasks.

[0088] Figure 4is a flowchart of a task processing method according to an embodiment of the present disclosure. Figure 4 As shown, the method comprises the following steps:

[0089] S401, obtaining a log of a target processing tool executing a target processing instruction, evaluation information, and evaluation indicators of the target processing tool.

[0090] In the disclosed embodiment, the processing tools may be evaluated, and the priorities of the processing tools may be sorted and optimized according to the evaluations of the processing tools.

[0091] Optionally, when evaluating processing tools, multi-dimensional data information needs to be taken into account. For example, the log of each target processing tool executing the target processing instruction may include the task completion effect of the processing tool (such as task completion time, accuracy, resource consumption, such as computing resources, memory usage, etc.), usage frequency (statistics of the number of times each processing tool is called), and internal performance indicators (such as stability and error rate during tool execution).

[0092] In some embodiments, when evaluating processing tools, external evaluation information also needs to be considered, and user feedback corresponding to each target processing tool can be collected as evaluation information (user satisfaction with task processing results, usability feedback, improvement suggestions, etc. are collected through a user evaluation system).

[0093] S402, calling a preset evaluation model, analyzing the log and evaluation information based on the evaluation index of the target processing tool, and obtaining a candidate evaluation value of the target processing tool under each evaluation index.

[0094] Based on the above data, comprehensive evaluation indicators are set to quantify the performance and quality of the processing tools.

[0095] S403, weighting the candidate evaluation values ​​of the target processing tool to obtain a target evaluation value of the target processing tool.

[0096] Optionally, different evaluation indicators may be assigned different weights, and the candidate evaluation values ​​may be weighted based on the weights of the evaluation indicators to obtain a target evaluation value of the target processing tool.

[0097] S404, dynamically adjusting the priority of the target processing tool based on the target evaluation value of the target processing tool.

[0098] For example, although processing tool A has high accuracy, it has a slow processing speed; processing tool B has a fast processing speed but a slightly lower accuracy; processing tool C performs well on certain types of text, but is not effective on other types. Based on the comprehensive evaluation indicators, the system re-ranks tool C to increase the priority of tool C in specific task types, while tools A and B adjust their positions based on overall performance.

[0099] S405: In response to the existence of a to-be-optimized processing tool whose target evaluation value is less than a preset evaluation threshold value among the target processing tools, the to-be-optimized processing tool and the association relationship between the to-be-optimized tools are deleted.

[0100] In some embodiments, if the target evaluation value of processing tool A is less than a preset evaluation threshold, processing tool A is marked as a processing tool to be optimized, and the processing tool to be optimized and the association relationship between the processing tool to be optimized are deleted to further improve the accuracy of task processing.

[0101] In the disclosed embodiment, a preset evaluation model is called, the log is analyzed based on the evaluation index of the target processing tool, the candidate evaluation values ​​of the target processing tool under each evaluation index are obtained, the candidate evaluation values ​​of the target processing tool are weighted, the target evaluation value of the target processing tool is obtained, and the priority of each target processing tool is dynamically adjusted based on the target evaluation value of each target processing tool. This can improve the adaptability and intelligence level of task processing. In response to the presence of a processing tool to be optimized whose target evaluation value is less than a preset evaluation threshold among the target processing tools, the processing tool to be optimized and the association relationship between the processing tools to be optimized are deleted. Through dynamic feedback processing, the task execution efficiency and accuracy can be improved, and the processing tools can be continuously optimized.

[0102] Figure 5 is a flowchart of a task processing method according to an embodiment of the present disclosure, such as Figure 5 As shown, the method is executed by a task processing system, including: receiving a task instruction, and calling a preset large language model to decompose the task instruction to obtain at least one target processing instruction; obtaining the association relationship between the reference processing instruction and the reference processing tool in the preset database; retrieving the target processing instruction based on the association relationship, and determining the target processing tool corresponding to the target processing instruction; for the target processing instruction for which the target processing tool is not retrieved, matching the target processing instruction based on the attribute information of each reference processing tool to determine whether there is a target processing tool corresponding to the target processing instruction in the reference processing tool, updating the association relationship for the target processing instruction matched with the target processing tool, and constructing the corresponding target processing tool for the target processing instruction that is not matched with the target processing tool, storing the constructed processing tool in the tool library, and updating the association relationship; calling the target processing tool corresponding to each target processing instruction from the preset tool library, and executing the target processing instruction based on the decomposition order of each target processing instruction to obtain the target processing result corresponding to the task instruction.

[0103] In some embodiments, for each target processing tool, a log, evaluation information, and evaluation indicators of the target processing tool executing the target processing instruction are obtained; a preset evaluation model is called, and the log and evaluation information are analyzed based on the evaluation indicators of the target processing tool to obtain candidate evaluation values ​​of the target processing tool under each evaluation indicator; the candidate evaluation values ​​of the target processing tool are weighted to obtain the target evaluation value of the target processing tool; based on the target evaluation value of each target processing tool, the priority of calling each target processing tool is dynamically adjusted, and if there is a processing tool to be optimized among the target processing tools whose target evaluation value is less than a preset evaluation threshold, the processing tool to be optimized and the association relationship between the tools to be optimized are deleted to update the association relationship and the tool library processing tool.

[0104] The task processing system can flexibly assemble subtasks and efficiently execute task instructions. The tool library supports continuous learning and updating of skills and improves the ability to cope with new tasks. Through dynamic feedback processing, the system can continuously optimize action plans and improve task execution efficiency and accuracy.

[0105] Figure 6 is a structural diagram of a task processing device according to an embodiment of the present disclosure, such as Figure 6 As shown, the task processing device 600 includes:

[0106] A first acquisition module 610 is used to receive a task instruction and call a preset large language model to decompose the task instruction to obtain at least one target processing instruction;

[0107] A second acquisition module 620 is used to acquire the association relationship between the reference processing instruction and the reference processing tool in the preset database;

[0108] A determination module 630 is configured to, in response to the existence of a first processing instruction belonging to a reference processing instruction in the target processing instruction, retrieve the first processing instruction based on the association relationship, and determine a target processing tool corresponding to the first processing instruction;

[0109] A first processing module 640 is used for constructing a target processing tool corresponding to the second processing instruction in response to the presence of a second processing instruction in the target processing instruction that does not belong to the reference processing instruction;

[0110] The second processing module 650 is used to call the target processing tool corresponding to the target processing instruction, execute the target processing instruction, and obtain the target processing result corresponding to the task instruction.

[0111] In some embodiments, the determination module 630 is further used to: obtain attribute information of a reference processing tool, the attribute information including a tool name, an application scope, and a calling method; traverse the second processing instructions, and for the third processing instruction currently traversed, match the third processing instruction based on the attribute information of the reference processing tool to determine whether there is a target processing tool corresponding to the third processing instruction in the reference processing tool; in response to the presence of a target processing tool corresponding to the third processing instruction in the reference processing tool, clear the third processing instruction from the second processing instruction.

[0112] In some embodiments, the determination module 630 is also used to: perform intent recognition on the third processing instruction to determine the first intent of the third processing instruction; perform intent recognition on the reference processing tool based on the application scope and calling method of the reference processing tool to determine the second intent of the reference processing tool; match the first intent and the second intent to obtain a first matching degree of the reference processing tool; match the third processing instruction with the tool name of the reference processing tool to obtain a second matching degree of the reference processing tool; and determine whether there is a target processing tool corresponding to the third processing instruction in the reference processing tool based on the first matching degree and the second matching degree of the reference processing tool and a preset priority of the reference processing tool.

[0113] In some implementations, the determination module 630 is further configured to: establish an association relationship between the third processing instruction and a target processing tool corresponding to the third processing instruction.

[0114] In some embodiments, the first processing module 640 is further used to: call the large language model to analyze the second processing instruction and the preset prompt word to generate candidate code; preset the deployment tool under the preset business environment to package and deploy the candidate code; call the debugging interface to debug the candidate code to obtain the target processing tool corresponding to the second processing instruction.

[0115] In some implementations, the first processing module 640 is further used to: store the target processing tool corresponding to the second processing instruction; and establish an association relationship between the second processing instruction and the target processing tool corresponding to the second processing instruction.

[0116] In some embodiments, there are two or more target processing instructions, and the second processing module 650 is also used to: obtain the decomposition order of the target processing instructions; call the target processing tool corresponding to the target processing instruction based on the decomposition order, execute the target processing instruction, and generate candidate processing results of the target processing instruction; integrate the candidate processing results of the target processing instruction based on the decomposition order to obtain the target processing result.

[0117] In some embodiments, the second processing module 650 is also used to: obtain a log of the target processing tool executing the target processing instruction and an evaluation index of the target processing tool; call a preset evaluation model, analyze the log based on the evaluation index of the target processing tool, and obtain candidate evaluation values ​​of the target processing tool under each evaluation index; weight the candidate evaluation values ​​of the target processing tool to obtain a target evaluation value of the target processing tool; and dynamically adjust the priority of the target processing tool based on the target evaluation value of the target processing tool.

[0118] In some embodiments, the second processing module 650 is further used to: in response to the existence of a to-be-optimized processing tool whose target evaluation value is less than a preset evaluation threshold among the target processing tools, delete the to-be-optimized processing tool and the association relationship between the to-be-optimized tools.

[0119] In the embodiment of the present disclosure, a preset large language model is called to decompose the task instruction to obtain at least one target processing instruction, the association relationship between the reference processing instruction and the reference processing tool in the preset database is obtained, the first processing instruction is retrieved based on the association relationship, and the target processing tool corresponding to the first processing instruction is determined, which can improve the efficiency of task processing, and the target processing tool corresponding to the second processing instruction is constructed, which can improve the accuracy of the newly appeared task processing instruction; the target processing tool corresponding to the target processing instruction is called, and the target processing instruction is executed to obtain the target processing result corresponding to the task instruction. The present disclosure can flexibly organize and coordinate the use of multiple tools, improve the ability of cross-task generalization, adapt to the changes of new tasks and new business environments, and complete complex tasks.

[0120] Based on the same application concept, the present disclosure also provides a task processing system, including the task processing device according to the above present disclosure.

[0121] The acquisition, transmission, storage, use, and processing of data in the technical solution disclosed in this disclosure are in compliance with the relevant provisions of national laws and regulations and do not violate public order and good morals. It should be noted that in the embodiments of this disclosure, some existing solutions in the industry such as certain software, components, and models may be mentioned, which should be considered as exemplary, and their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.

[0122] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product. Figure 7It is a block diagram of an electronic device for implementing an embodiment of the present disclosure. The electronic device can implement the task processing method of the embodiment of the present disclosure, and the electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0123] like Figure 7 As shown, the device 700 includes a computing unit 701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 702 or a computer program loaded from a storage unit 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the device 700 can also be stored. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0124] A number of components in the device 700 are connected to the I / O interface 705, including: an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a disk, an optical disk, etc.; and a communication unit 709, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 709 allows the device 700 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0125] The computing unit 701 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 701 performs the various methods and processes described above, such as the task processing method. For example, in some embodiments, the task processing method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the task processing method described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform the task processing method in any other appropriate manner (e.g., by means of firmware).

[0126] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0127] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0128] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0129] To provide interaction with a target object, the systems and techniques described herein may be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the target object; and a keyboard and pointing device (e.g., a mouse or trackball) through which the target object can provide input to the computer. Other types of devices may also be used to provide interaction with the target object; for example, the feedback provided to the target object may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the target object may be received in any form (including acoustic input, voice input, or tactile input).

[0130] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a target object computer with a graphical target object interface or a web browser through which the target object can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0131] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0132] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0133] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A task processing method, wherein: include: Receive a task instruction, and call a preset large language model to decompose the task instruction to obtain at least one target processing instruction; Obtaining the association relationship between the reference processing instruction and the reference processing tool in the preset database; In response to the presence of a first processing instruction belonging to the reference processing instruction in the target processing instructions, searching the first processing instruction based on the association relationship to determine a target processing tool corresponding to the first processing instruction; In response to the presence of a second processing instruction in the target processing instruction that does not belong to the reference processing instruction, constructing the target processing tool corresponding to the second processing instruction; The target processing tool corresponding to the target processing instruction is called to execute the target processing instruction to obtain the target processing result corresponding to the task instruction.

2. The method according to claim 1, wherein: Before constructing the target processing tool corresponding to the second processing instruction, the method further includes: Acquire attribute information of the reference processing tool, wherein the attribute information includes a tool name, an application scope, and a calling method; Traversing the second processing instructions, and for the third processing instruction currently traversed, matching the third processing instruction based on the attribute information of the reference processing tool to determine whether the target processing tool corresponding to the third processing instruction exists in the reference processing tool; In response to the presence of the target processing tool corresponding to the third processing instruction in the reference processing tool, the third processing instruction is cleared from the second processing instructions.

3. The method according to claim 2, wherein: The matching the third processing instruction based on the attribute information of the reference processing tool to determine whether the target processing tool corresponding to the third processing instruction exists in the reference processing tool includes: Performing intent recognition on the third processing instruction to determine a first intent of the third processing instruction; Based on the application scope and the calling method of the reference processing tool, performing intent recognition on the reference processing tool to determine a second intent of the reference processing tool; Matching the first intention with the second intention to obtain a first matching degree of the reference processing tool; Matching the third processing instruction with the tool name of the reference processing tool to obtain a second matching degree of the reference processing tool; Based on the first matching degree and the second matching degree of the reference processing tool and a preset priority of the reference processing tool, it is determined whether the target processing tool corresponding to the third processing instruction exists in the reference processing tool.

4. The method according to claim 3, wherein: Before clearing the third processing instruction from the second processing instruction, the method further includes: An association relationship is established between the third processing instruction and the target processing tool corresponding to the third processing instruction.

5. The method according to any one of claims 1 to 4, wherein: The constructing the target processing tool corresponding to the second processing instruction includes: Calling the large language model to analyze the second processing instruction and the preset prompt word to generate a candidate code; Pre-set deployment tools in a preset business environment to package and deploy the candidate codes; The debugging interface is called to debug the candidate code to obtain the target processing tool corresponding to the second processing instruction.

6. The method according to claim 5, wherein: After constructing the target processing tool corresponding to the second processing instruction, the method further includes: storing the target processing tool corresponding to the second processing instruction; An association relationship is established between the second processing instruction and the target processing tool corresponding to the second processing instruction.

7. The method according to claim 3 or 4, wherein: The target processing instructions are two or more, and the calling of the target processing tool corresponding to the target processing instruction and the execution of the target processing instruction to obtain the target processing result corresponding to the task instruction includes: Obtaining a decomposition order of the target processing instruction; calling the target processing tool corresponding to the target processing instruction based on the decomposition order, executing the target processing instruction, and generating a candidate processing result of the target processing instruction; The candidate processing results of the target processing instruction are integrated based on the decomposition order to obtain the target processing result.

8. The method according to claim 7, wherein: After integrating the candidate processing results of the target processing instruction based on the decomposition order to obtain the target processing result, the method further includes: Obtaining a log of the target processing tool executing the target processing instruction, evaluation information, and an evaluation index of the target processing tool; Calling a preset evaluation model, analyzing the log and the evaluation information based on the evaluation index of the target processing tool, and obtaining a candidate evaluation value of the target processing tool under each evaluation index; Weighting the candidate evaluation values ​​of the target processing tool to obtain a target evaluation value of the target processing tool; Based on the target evaluation value of the target processing tool, the priority of the target processing tool is dynamically adjusted.

9. The method according to claim 8, wherein: Also includes: In response to the existence of a to-be-optimized processing tool whose target evaluation value is less than a preset evaluation threshold among the target processing tools, the to-be-optimized processing tool and the association relationship between the to-be-optimized tools are deleted.

10. A task processing device, wherein: include: A first acquisition module is used to receive a task instruction and call a preset large language model to decompose the task instruction to obtain at least one target processing instruction; A second acquisition module is used to acquire the association relationship between the reference processing instruction and the reference processing tool in the preset database; a determination module, configured to, in response to the existence of a first processing instruction belonging to the reference processing instruction in the target processing instruction, retrieve the first processing instruction based on the association relationship, and determine a target processing tool corresponding to the first processing instruction; A first processing module, configured to construct the target processing tool corresponding to the second processing instruction in response to the presence of a second processing instruction in the target processing instruction that does not belong to the reference processing instruction; The second processing module is used to call the target processing tool corresponding to the target processing instruction, execute the target processing instruction, and obtain the target processing result corresponding to the task instruction.

11. The device according to claim 10, wherein: The determining module is further used for: Acquire attribute information of the reference processing tool, wherein the attribute information includes a tool name, an application scope, and a calling method; Traversing the second processing instructions, and for the third processing instruction currently traversed, matching the third processing instruction based on the attribute information of the reference processing tool to determine whether the target processing tool corresponding to the third processing instruction exists in the reference processing tool; In response to the presence of the target processing tool corresponding to the third processing instruction in the reference processing tool, the third processing instruction is cleared from the second processing instructions.

12. The device according to claim 11, wherein The determining module is further used for: Performing intent recognition on the third processing instruction to determine a first intent of the third processing instruction; Based on the application scope and the calling method of the reference processing tool, performing intent recognition on the reference processing tool to determine a second intent of the reference processing tool; Matching the first intention with the second intention to obtain a first matching degree of the reference processing tool; Matching the third processing instruction with the tool name of the reference processing tool to obtain a second matching degree of the reference processing tool; Based on the first matching degree and the second matching degree of the reference processing tool and a preset priority of the reference processing tool, it is determined whether the target processing tool corresponding to the third processing instruction exists in the reference processing tool.

13. The device according to claim 12, wherein: The determining module is further used for: An association relationship is established between the third processing instruction and the target processing tool corresponding to the third processing instruction.

14. The device according to any one of claims 10 to 13, wherein: The first processing module is further used for: Calling the large language model to analyze the second processing instruction and the preset prompt word to generate a candidate code; Pre-set deployment tools in a preset business environment to package and deploy the candidate codes; The debugging interface is called to debug the candidate code to obtain the target processing tool corresponding to the second processing instruction.

15. The device according to claim 14, wherein: The first processing module is further used for: storing the target processing tool corresponding to the second processing instruction; An association relationship is established between the second processing instruction and the target processing tool corresponding to the second processing instruction.

16. The device according to claim 12 or 13, wherein: The target processing instructions are two or more, and the second processing module is further used for: Obtaining a decomposition order of the target processing instruction; calling the target processing tool corresponding to the target processing instruction based on the decomposition order, executing the target processing instruction, and generating a candidate processing result of the target processing instruction; The candidate processing results of the target processing instruction are integrated based on the decomposition order to obtain the target processing result.

17. The device according to claim 16, wherein: The second processing module is further used for: Obtaining a log of the target processing tool executing the target processing instruction and an evaluation index of the target processing tool; Calling a preset evaluation model, analyzing the log based on the evaluation index of the target processing tool, and obtaining a candidate evaluation value of the target processing tool under each evaluation index; Weighting the candidate evaluation values ​​of the target processing tool to obtain a target evaluation value of the target processing tool; Based on the target evaluation value of the target processing tool, the priority of the target processing tool is dynamically adjusted.

18. The device according to claim 17, wherein: The second processing module is further used for: In response to the existence of a to-be-optimized processing tool whose target evaluation value is less than a preset evaluation threshold among the target processing tools, the to-be-optimized processing tool and the association relationship between the to-be-optimized tools are deleted.

19. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method according to any one of claims 1 to 9.

20. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-9.

21. A computer program product comprising a computer program, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 9.

22. A task processing system, comprising the device according to any one of claims 10-18.

Images