Task processing method and device, electronic equipment and storage medium
By dividing the task processing process into thought chain skills and atomic skills, the professional and standardized processing of content marketing tasks is achieved, improving the reliability and automation of task processing and solving the efficiency and reliability problems of complex tasks with multiple stages.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- OBJECT INTEGRITY (SHANGHAI) TECHNOLOGY CO LTD
- Filing Date
- 2025-12-31
- Publication Date
- 2026-05-12
AI Technical Summary
When dealing with complex content marketing tasks involving multiple stages, intelligent agents struggle to efficiently and reliably complete various stages such as data acquisition, cleaning, analysis, and writing, resulting in low task processing efficiency and poor result reliability.
By dividing the task processing process into thought chain skills and atomic skills, the thought chain skills encapsulate solutions, and the atomic skills encapsulate executable logic code. The execution flow is orchestrated and the atomic skills are called to execute the steps, thereby achieving professional and standardized task processing.
It improves the reliability and automation of task processing results, reduces human intervention, and increases processing efficiency.
Smart Images
Figure CN122021698A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence technology, and in particular to a task processing method, apparatus, electronic device, and storage medium. Background Technology
[0002] With the rapid advancement of Large Language Model (LLM) technology, intelligent agents, with their capabilities of autonomous perception, decision-making, and execution, are increasingly being applied to diverse business scenarios to intelligently handle various tasks within those scenarios.
[0003] Taking content marketing as an example, intelligent agents can handle some aspects of the work intelligently, such as generating campaign copy, writing product descriptions, and providing topic suggestions. While these agents can utilize a range of available tools during task processing, when dealing with complex, multi-stage content marketing tasks, such as market analysis requiring data acquisition, cleaning, analysis, and document creation, the agent must not only understand the task objectives but also select and combine tools. This process is highly susceptible to errors, leading to low efficiency and unreliable results. Summary of the Invention
[0004] In view of this, the embodiments of this application provide a task processing method that can achieve efficient processing of complex tasks with multiple stages and improve the reliability of task processing results.
[0005] In a first aspect, embodiments of this application provide a task processing method, the method comprising: receiving a target task processing request input by a user; determining a thought chain skill adapted to the target task based on the target task processing request, the thought chain skill describing a solution to the target task; using the thought chain skill to orchestrate an execution flow of the target task based on the solution described by the thought chain skill, the execution flow including at least one step requiring execution using an atomic skill, wherein the atomic skill encapsulates executable logic code for implementing a predetermined function, and when the atomic skill is used, the executable logic code is invoked to implement the predetermined function; and executing the steps using the atomic skill according to the orchestrated execution flow to obtain a processing result of the target task.
[0006] According to one embodiment of this application, the thought chain skill includes a thought chain description, which includes at least one of a checklist, a step sequence, branching conditions, and feedback loop information, wherein: the checklist describes the content that needs to be checked at the checkpoints in the solution or the input information that needs to be obtained in the solution; the step sequence describes the execution steps in the solution; the branching conditions describe the logical judgment conditions for different execution paths in the solution; and the feedback loop information describes the processing principles for predetermined events.
[0007] According to one embodiment of this application, the atomic skills include data retrieval skills, table statistical analysis skills, report generation skills, report generation skills, image generation skills, and video script storyboard generation skills.
[0008] According to one embodiment of this application, determining the mind chain skill adapted to the target task based on the target task processing request includes: dividing the target task into at least two task stages based on the target task processing request; determining the mind chain skill adapted to each task stage for each task stage; arranging the execution flow of the target task using the mind chain skill to create a solution based on the mind chain skill description includes: arranging the execution flow of the task stage using the mind chain skill adapted to the task stage, wherein the execution flow of the task stage includes at least one step that needs to be executed using an atomic skill; executing the step using the atomic skill according to the arranged execution flow to obtain the processing result of the target task includes: for each task stage, executing the step using the atomic skill according to the arranged execution flow of the task stage to obtain the execution result of each task stage; and obtaining the execution result of the target task based on the execution result of each task stage.
[0009] According to one embodiment of this application, the task phase includes a first task phase and a second task phase executed sequentially; after obtaining the execution result of the first task phase, the method further includes: evaluating whether the execution result of the first task phase meets a predetermined quality condition; if the execution result of the first task phase meets the predetermined quality condition, performing the steps using the atomic skills according to the arranged execution flow of the second task phase to obtain the execution result of the second task phase; if the execution result of the first task phase does not meet the predetermined quality condition, determining the reason why the execution result does not meet the predetermined quality condition, and determining a subsequent execution path based on the reason to adjust the execution result of the first task phase.
[0010] According to one embodiment of this application, the subsequent execution path includes: adjusting the execution flow of the orchestrated first task stage and then re-executing the adjusted execution flow of the first task stage; rearranging the execution flow of the first task stage and then executing the rearranged execution flow of the first task stage; adjusting the input data for the orchestrated execution flow of the first task stage and then re-executing the orchestrated execution flow of the first task stage.
[0011] According to one embodiment of this application, determining the mind chain skill adapted to the target task based on the target task processing request includes: providing the target task processing request to the mind chain skill agent, so that the mind chain agent determines the mind chain skill adapted to the target task based on the task processing request, and then uses the mind chain skill to arrange the execution flow of the target task with a solution described by the mind chain skill.
[0012] According to one embodiment of this application, the step of using the atomic skill to perform the steps according to the arranged execution flow includes: receiving the arranged execution flow provided by the mind chain skill agent, and providing the arranged execution flow to the atomic skill agent so that the atomic skill agent performs the steps according to the arranged execution flow; or, providing the arranged execution flow to the atomic skill agent through the mind chain agent so that the atomic skill agent performs the steps according to the arranged execution flow.
[0013] According to one embodiment of this application, the target task is a content marketing task, which includes an industry insight task, a competitor research task, a content planning task, a creator matching task, an advertising placement task, and a marketing effect review task.
[0014] Secondly, embodiments of this application also provide a task processing apparatus, the apparatus comprising: a data receiving module for receiving a target task processing request input by a user; a first processing module for determining a thought chain skill adapted to the target task based on the target task processing request, the thought chain skill describing a solution to the target task; a second processing module for using the thought chain skill to orchestrate an execution flow of the target task based on the solution described by the thought chain skill, the execution flow including at least one step requiring execution using an atomic skill, wherein the atomic skill encapsulates executable logic code for implementing a predetermined function, and the executable logic code is invoked when the atomic skill is used to implement the predetermined function; and a first execution module for executing the steps using the atomic skill according to the orchestrated execution flow to obtain a processing result of the target task.
[0015] According to one embodiment of this application, the thought chain skill includes a thought chain description, which includes at least one of a checklist, a step sequence, branching conditions, and feedback loop information, wherein: the checklist describes the content that needs to be checked at the checkpoints in the solution or the input information that needs to be obtained in the solution; the step sequence describes the execution steps in the solution; the branching conditions describe the logical judgment conditions for different execution paths in the solution; and the feedback loop information describes the processing principles for predetermined events.
[0016] According to one embodiment of this application, the atomic skills include data retrieval skills, table statistical analysis skills, report generation skills, report generation skills, image generation skills, and video script storyboard generation skills.
[0017] According to one embodiment of this application, the first processing module is specifically configured to: divide the target task into at least two task stages based on the target task processing request; and determine a thought chain skill adapted to each task stage for each task stage; the second processing module is specifically configured to: use the thought chain skill adapted to the task stage to arrange the execution flow of the task stage, wherein the execution flow of the task stage includes at least one step that needs to be executed using an atomic skill; the first execution module is specifically configured to: for each task stage, execute the steps using the atomic skill according to the arranged execution flow of the task stage to obtain the execution result of each task stage; and obtain the execution result of the target task based on the execution result of each task stage.
[0018] According to one embodiment of this application, the task phase includes a first task phase and a second task phase executed sequentially; the apparatus further includes: a decision module, configured to, after obtaining the execution result of the first task phase, evaluate whether the execution result of the first task phase meets a predetermined quality condition; if the execution result of the first task phase does not meet the predetermined quality condition, determine the reason why the execution result does not meet the predetermined quality condition, and determine a subsequent execution path based on the reason to adjust the execution result of the first task phase; the first execution module is specifically configured to, if the execution result of the first task phase meets the predetermined quality condition, execute the steps using the atomic skill according to the arranged execution flow of the second task phase to obtain the execution result of the second task phase.
[0019] According to one embodiment of this application, the subsequent execution path includes: adjusting the execution flow of the orchestrated first task stage and then re-executing the adjusted execution flow of the first task stage; rearranging the execution flow of the first task stage and then executing the rearranged execution flow of the first task stage; adjusting the input data of the orchestrated execution flow of the first task stage and then re-executing the orchestrated execution flow of the first task stage.
[0020] According to one embodiment of this application, the first processing module is further configured to provide the target task processing request to the thought chain skill agent, so that the thought chain agent determines the thought chain skill adapted to the target task based on the task processing request, and then uses the thought chain skill to arrange the execution flow of the target task with a solution based on the thought chain skill description.
[0021] According to one embodiment of this application, the first execution module is specifically configured to receive the orchestrated execution flow provided by the mind chain skill agent, and provide the orchestrated execution flow to the atomic skill agent, so that the atomic skill agent performs the steps using the atomic skills according to the orchestrated execution flow; or, the mind chain agent provides the orchestrated execution flow to the atomic skill agent, so that the atomic skill agent performs the steps using the atomic skills according to the orchestrated execution flow.
[0022] According to one embodiment of this application, the target task is a content marketing task, which includes an industry insight task, a competitor research task, a content planning task, a creator matching task, an advertising placement task, and a marketing effect review task.
[0023] Thirdly, embodiments of this application also provide an electronic device, the electronic device comprising: a processor and a memory, wherein the memory is used to store executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, for executing any of the task processing methods described in the first aspect above.
[0024] Fourthly, embodiments of this application also provide a computer-readable storage medium storing one or more programs, which can be executed by one or more processors to implement the task processing method described in any of the first aspects above.
[0025] This application provides a task processing method, apparatus, system, electronic device, and storage medium. Skills are categorized into two types: thought chains and atoms. Thought chain skills encapsulate professional solutions, while atomic skills encapsulate executable logic code for various capabilities. During task processing, thought chain skills are responsible for task planning and arranging the task execution flow based on the encapsulated solutions. Atomic skills are responsible for executing this execution flow. This approach specializes task solutions and standardizes the task processing process, effectively improving the reliability of task processing results. Furthermore, the use of thought chain and atomic skills reduces human intervention in the task processing process, increasing automation and efficiency. Attached Figure Description
[0026] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0027] Figure 1 This is a schematic flowchart of a task processing method provided in an embodiment of this application; Figure 2 A flowchart illustrating step S120 in a task processing method provided in an embodiment of this application; Figure 3 A flowchart illustrating step S140 in a task processing method provided in an embodiment of this application; Figure 4 This is a schematic diagram of a task processing device architecture provided in an embodiment of this application; Figure 5 This is a schematic block diagram illustrating the architecture of an embodiment of the electronic device of this application. Detailed Implementation
[0028] The embodiments of this application will now be described in detail with reference to the accompanying drawings.
[0029] It should be understood that the described embodiments are merely some, not all, of the embodiments in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.
[0030] To facilitate a better understanding by those skilled in the art of the task processing safeguards provided in the embodiments of the present invention, the technical terms in the embodiments of the present invention will first be explained: Large Language Model (LLM): A deep learning model that has been trained on a large corpus and is capable of understanding and generating human natural language, and possesses certain reasoning and planning abilities.
[0031] Large Language Model Agent: refers to a software entity built on a large language model that has the ability to autonomously understand tasks, plan steps, invoke tools or skills, and produce results.
[0032] Operating-System-Level Tool Agent: refers to a large language model agent that runs in a controlled computing environment and can represent other agents to execute terminal commands, read and write files, call scripts or external tools, and can serve as the execution subject of this application.
[0033] Skill: refers to a standardized, modular unit of intelligent agent capability encapsulation. It teaches intelligent agents to execute specific workflows by encapsulating the instructions, scripts, and resource organization required to complete a specific task into a standardized folder.
[0034] Atomic Skill: A skill proposed in this embodiment of the invention, which is a modality of skill, refers to a single-function, logically defined capability unit driven by code execution. It usually corresponds to the underlying programmable tool capability and is typically implemented through one or more terminal commands, script calls, or API requests.
[0035] ThinkFlow Skill / Chain-of-Thought Skill: Another skill proposed in this embodiment of the invention, which is another modality of skill, refers to the encapsulation of task solutions, such as the explicit encapsulation of expert implicit cognitive models (Mental Model), describing the solution to the task, responsible for intent recognition, task decomposition and logical reasoning, which can rely on atomic skills, and can arrange multiple atomic skills in temporal and logical order to solve complex and highly uncertain tasks.
[0036] See Figure 1 As shown, this application uses a content marketing scenario as an example to illustrate a task processing method provided by this application, including the following steps: S110, Receive the target task processing request input by the user.
[0037] In this step, the user's target task processing request is received through at least one of the following methods: a text input box in the graphical user interface, a voice input box, a file upload interface, or an API interface integrated with a third-party application. The target task processing request can be natural language text, a structured query statement such as SQL, a pre-formatted command code, or a file containing metadata. The request explicitly or implicitly includes a task description of the target task, such as "generate a third-quarter sales report," and output requirements, such as "generate a PDF document."
[0038] S120. Based on the target task processing request, determine the thought chain skill that is compatible with the target task, wherein the thought chain skill describes the solution to the target task.
[0039] In this step, after receiving the target task processing request, the request can first be preprocessed, such as semantic error correction. Then, the preprocessed target task processing request is subjected to intent understanding and target parsing to determine the mind chain skill that is suitable for the target task. This mind chain skill describes the solution to the target task.
[0040] In some embodiments of the present invention, the mind chain skills proposed in these embodiments may include: The entry file includes a functional description, a thought process description, and routing rules. Referenced files; The functional description indicates the problem that the target skill solves, the thought chain description describes the solution to the problem, the referenced files include the files that need to be called in the solution described in the thought chain description, and the routing rules indicate the conditions for calling the referenced documents, that is, under what conditions which referenced files are called.
[0041] Specifically, the entry file can be the SKILL.md document.
[0042] For example, in the entry file, the thought chain description may include at least one of a checklist, a sequence of steps, branching conditions, and feedback loop information, which describe the solution to the problem. The checklist describes what needs to be checked at each checkpoint in the solution, such as input files required before the solution or a step within it can run. The sequence of steps describes the execution steps in the solution. Branching conditions describe the logical judgment conditions for different execution paths in the solution, guiding the solution's flow towards different execution paths. Feedback loop information describes the handling principles for predetermined events, such as the execution result of a step or the handling principles for external feedback, such as how to handle user feedback received during solution execution.
[0043] In this embodiment, different thought chain skills can describe solutions to the same or different tasks. That is, different thought chain skills can encapsulate the thought chain descriptions for different tasks, including checklists, step sequences, feedback loop information, decision branch information, etc. Among them, the checklist indicates the input information and execution process that needs to be obtained in the solution, such as key steps, elements or precautions in the task execution process. For example, for a competitor research task, the checklist may include: a list of competitors and information about the competitors. The step sequence clearly indicates all the execution steps and processes of the solution. For example, first, data is collected, then data analysis is performed, and finally the analysis results and decisions are output. The feedback loop information indicates the handling principles for predetermined events. The predetermined events may be events such as step execution errors, execution results not meeting preset conditions, or user feedback information. The feedback loop information indicates what should be done for the predetermined event.
[0044] S130. Using the thought chain skill to orchestrate the execution flow of the target task based on the solution described by the thought chain skill, the execution flow includes at least one step that needs to be executed using an atomic skill, wherein the atomic skill encapsulates executable logic code for implementing a predetermined function, and when the atomic skill is used, the executable logic code is called to implement the predetermined function.
[0045] In this step, the target task is planned based on the identified mind chain skills that are suitable for it, thereby planning the execution flow of the target task. If the target task is very complex and includes multiple task stages, all task stages of the entire target task can be planned first, and for each task stage, the execution flow of the target task can be arranged based on the solutions encapsulated in the mind chain skills.
[0046] The execution process includes at least one step that requires the use of atomic skills. The atomic skills encapsulate executable logic code for implementing a predetermined function. When the atomic skills are used, the executable logic code is called to implement the predetermined function.
[0047] In some embodiments of the present invention, the atomic skills proposed in these embodiments may include: An entry file, which includes a function description that indicates the function of an atomic skill; Solution documentation, which stores executable logic code that implements a predetermined function, may be a script file stored in the script file directory ( / scripts).
[0048] In some embodiments, the atomic skills include data retrieval skills, table statistical analysis skills, report generation skills, report generation skills, image generation skills, video script storyboard generation skills, etc., that is, the functions encapsulated by the atomic skills are the corresponding data retrieval, table statistical analysis, report generation, report generation, image generation, video script storyboard generation, etc.
[0049] In this embodiment, each atomic intelligence encapsulates a type of executable logic code with relatively simple functions and limited operation steps to achieve a predetermined function, such as data retrieval skills, table statistical analysis skills, report generation skills, image generation skills, and video script storyboard generation skills. When using an atomic skill, the executable logic code is invoked to achieve the predetermined function. For example, the table statistical analysis skill reads the input table file by invoking the executable code, then performs specific statistical calculations, and finally generates a new result table.
[0050] Furthermore, atomic skills can be reused by different mind chain skills to adapt to different task processing scenarios. For example, data retrieval skills can be used in industry insight tasks (called by the industry insight mind chain) or in competitor research tasks (called by the competitor research mind chain). By combining and expanding mind chain skills, more task processing scenarios can be adapted.
[0051] In some embodiments, when an atomic skill calls executable logic code to implement a predetermined function, it can also, according to a predefined error detection mechanism and a corresponding fallback strategy, call the corresponding fallback strategy to continue implementing the predetermined function when a predefined error is detected. The predefined error detection mechanism includes: network failure, insufficient permissions, data format errors, etc.; the corresponding fallback strategy includes retrying the call to executable code a preset number of times, switching to an alternative data source, and returning structured data with error codes and prompts.
[0052] S140. According to the arranged execution flow, the atomic skills are used to perform the steps to obtain the processing result of the target task.
[0053] In this step, following all the execution processes arranged by the mind chain skills, the atomic skills are called sequentially to execute, the results of each process are obtained and summarized, and the final processing result of the target task is obtained and presented to the user.
[0054] The task processing method provided in this application divides skills into atomic skills and thought chain skills. Thought chain skills encapsulate solutions, while atomic skills encapsulate executable logic code for various capabilities. During task processing, thought chain skills plan the task and, based on the encapsulated solutions, orchestrate the task's execution flow. Atomic skills are responsible for executing this flow. This approach specializes the task-specific solutions and standardizes the task processing process, effectively improving the reliability of task processing results. Furthermore, the use of thought chain skills and atomic skills reduces human intervention, increasing the automation and efficiency of the task processing process.
[0055] In some embodiments, see Figure 2 Step S120: Based on the target task processing request, determine the thought chain skills that are suitable for the target task, including: S121. Based on the target task processing request, the target task is divided into at least two task stages; S122. For each of the task stages, determine the mind chain skills that are suitable for the task stage.
[0056] In this embodiment, after obtaining the user's input of the target task processing request, semantic understanding and target parsing can be performed first, and the target task can be planned into at least two task stages. For each task stage, the corresponding thinking chain skills are determined, and the task processing process is standardized, which can effectively improve the reliability of the task processing results.
[0057] The process of orchestrating the execution flow of the target task using the mind chain skills to describe the solution includes: Using the thought chain skills adapted to the task phase, the execution flow of the task phase is arranged, and the execution flow of the task phase includes at least one step that requires the use of atomic skills; See Figure 3 Step S140, the step of performing the steps according to the arranged execution flow and using the atomic skills to obtain the processing result of the target task includes: S141. For each of the task stages, according to the arranged execution flow of the task stages, the atomic skills are used to execute the steps to obtain the execution result of each task stage; S142. Based on the execution results of each of the task stages, obtain the execution result of the target task.
[0058] In this embodiment, after obtaining the user's input request for processing the target task, semantic understanding and target parsing are performed. The target task is planned into at least two task stages. In each task stage, each execution flow of that stage is selected and orchestrated. Each execution flow includes at least one step that requires the use of atomic skills. According to the orchestrated task stages and execution flows, each step is executed sequentially using atomic skills, and the execution result of each task stage is obtained. The execution results obtained from each task stage are summarized to obtain the final execution result of the target task. By making the task stages, key steps, and decision branches explicit, the execution basis and results of each step can be traced, significantly improving the transparency and controllability of complex automated processes.
[0059] Specifically, taking a typical content marketing scenario as an example, the task processing of "Developing an Integrated Marketing Plan for a New Product Launch," after receiving the user's input request for the target task, "Developing an Integrated Marketing Plan for a New Product Launch," divides the overall target task into several task stages, such as industry insight tasks, competitor research tasks, content planning tasks, creator matching tasks, advertising placement tasks, and marketing effect review tasks. For each task stage, determine the corresponding mindset skills, such as industry insight mindset skills, competitor research mindset skills, content planning mindset skills, creator matching mindset skills, advertising placement mindset skills, and review and evaluation mindset skills. The process involves: using mind chain skills tailored to each task stage to orchestrate the execution flow for each stage, where each stage includes at least one step requiring the use of atomic skills; then, following the orchestrated task stages and execution flow, sequentially using atomic skills to execute each step and obtain the execution results for each stage. For example, in the industry and user insight stage, data retrieval skills can be used sequentially to obtain relevant content samples, table statistical analysis skills can be used to analyze user interaction data, and report generation skills can be used to generate preliminary insight documents for the industry and user insight stage; finally, the execution results obtained from each task stage are summarized to generate a complete content marketing plan.
[0060] In some embodiments, the task phase includes a first task phase and a second task phase executed sequentially, wherein the execution of the second task phase depends on the first task phase, and after obtaining the execution result of the first task phase, the method further includes: Evaluate whether the execution results of the first task phase meet the predetermined quality conditions; If the execution result of the first task phase meets the predetermined quality conditions, the steps are executed using the atomic skills according to the arranged execution flow of the second task phase to obtain the execution result of the second task phase; If the execution result of the first stage does not meet the predetermined quality conditions, determine the reason why the execution result does not meet the predetermined quality conditions, and determine the subsequent execution path based on the reason to adjust the execution result of the first task stage.
[0061] In this embodiment, during the execution of steps using atomic skills according to the arranged execution flow, the execution results generated after each atomic skill execution step can be evaluated, and the subsequent execution path can be dynamically adjusted based on the evaluation results.
[0062] Specifically, a complete target task includes a first task phase and a second task phase executed sequentially. After executing the first task phase according to the programmed execution flow and obtaining the execution results of the atomic skills in the first task phase, the execution results are compared with predetermined quality conditions. It is determined whether the execution results of the first task phase meet the predetermined quality conditions. If they do, the execution flow of the second task phase continues, using the corresponding atomic skills to execute each step of the second task phase. If the predetermined quality conditions are not met, the reasons for the failure to meet the predetermined quality conditions are further identified, and subsequent execution paths are dynamically adjusted based on these reasons. For example, if the research report output in the competitor research phase is blank or contains garbled text, and analysis determines that the reason for the blank or garbled text is insufficient competitor samples during the search, the mind chain skill can automatically expand the search scope or adjust the search keywords, and re-call the relevant atomic skills to execute this step, outputting the execution results of this phase.
[0063] In some embodiments, the subsequent execution path includes: After adjusting the execution flow of the first task phase in the arrangement, the adjusted execution flow of the first task phase is re-executed. After rearranging the execution flow of the first task phase, the rearranged execution flow of the first task phase is executed. After adjusting the input data of the first stage of the orchestrated execution flow, the execution flow of the first task stage of the orchestrated process is re-executed.
[0064] In this embodiment, the subsequent execution path can be dynamically adjusted based on the execution results generated after each atomic skill's execution step. When adjusting the subsequent execution path, the execution flow of the first task stage can be adjusted, for example, adjusting the execution steps of the atomic skills in the execution flow; the execution flow of the first task stage can be rearranged, for example, deleting the original atomic skills, selecting other atomic skills, and rearranging the steps to be executed by the atomic skills in the first stage's execution flow; the input data of the first stage's execution flow can also be adjusted, for example, in the data retrieval skill in the first stage, automatically expanding the search scope or adjusting the search keywords, and re-calling the atomic skill to execute the step.
[0065] In some embodiments, determining the thought chain skills adapted to the target task based on the target task processing request includes: The target task processing request is provided to the thought chain skill agent, so that the thought chain skill agent determines the thought chain skill that is suitable for the target task based on the task processing request, and then uses the thought chain skill to arrange the execution flow of the target task with a solution based on the thought chain skill description.
[0066] In this embodiment, multiple agents can collaborate during task processing. These agents can include terminal agents and thought chain skill agents. By abstracting and decomposing tasks in layers, the terminal agent is responsible for interacting with the user. After receiving a target task processing request, it provides the target task request to the thought chain agent. The thought chain skill agent plans and makes decisions on the target task, divides the target task into different task stages, determines the corresponding thought chain skill for each task stage, and arranges the execution process and steps of the atomic skills for each task stage.
[0067] In some embodiments, performing the steps using the atomic skills according to the orchestrated execution flow includes: The system receives the orchestrated execution flow provided by the thought chain skill agent, and provides the orchestrated execution flow to the atomic skill agent, so that the atomic skill agent executes the steps using the atomic skills according to the orchestrated execution flow; or, The thought chain agent provides the orchestrated execution flow to the atomic skill agent, so that the atomic skill agent can execute the steps using the atomic skills according to the orchestrated execution flow.
[0068] In this embodiment, multiple agents can collaborate during task processing. These agents may include a terminal tool agent, a thought chain skill agent, and an atomic skill agent. The terminal tool agent is responsible for interacting with the user. Upon receiving a target task processing request, it always provides the target task request to the thought chain skill agent. The thought chain skill agent plans and makes decisions regarding the target task, determines the appropriate thought chain skill, and arranges the execution process and steps of the atomic skills for processing the target task. After the thought chain skill agent arranges the execution flow, it provides the flow directly to the atomic skill agent or the terminal agent provides it to the atomic skill agent. The atomic skill agent then sequentially calls executable logic code according to the arranged steps to implement the corresponding function.
[0069] Specifically, the terminal tool intelligence agent is located at the bottom layer, responsible for interacting with the operating system and various tools. The atomic intelligence agent is located in the middle layer, responsible for encapsulating and combining executable logic code. This facilitates the upper-layer thinking chain skill agent to orchestrate and call it, and allows for direct invocation of executable code to implement response functions. The thinking chain intelligence agent is located at the top layer, responsible for planning and decision-making based on the target task, and orchestrating appropriate atomic skills at different task stages. Through the collaborative action of the terminal tool intelligence agent, atomic skill intelligence agent, and thinking chain intelligence agent, the task processing process is executed together, improving the automation level and processing efficiency of the task processing process.
[0070] In some embodiments, the scheduling of thought chain skill agents and atomic skill agents can be implemented based on message queues, event buses, function orchestration platforms, or workflow engines.
[0071] The embodiments of this invention can be applied to content marketing scenarios to handle content marketing tasks. Content marketing tasks are generally quite complex. The combination, reuse, and collaboration of atomic skills and thought chain skills enable the system to quickly adapt to different content marketing scenarios and tasks, achieving decomposable, programmable, and scalable processing of complex content marketing tasks, improving the automation level, business fit, maintainability, and explainability of the overall chain.
[0072] The target task is a content marketing task, which includes industry insight tasks, competitor research tasks, content planning tasks, creator matching tasks, advertising placement tasks, and marketing effect review tasks.
[0073] See Figure 4 This application also provides a task processing device, the device comprising: Data receiving module 41 is used to receive the target task processing request input by the user; The first processing module 42 is used to determine a thought chain skill that is compatible with the target task based on the target task processing request, wherein the thought chain skill describes the solution to the target task; The second processing module 43 is used to use the thought chain skill to arrange the execution flow of the target task based on the solution described by the thought chain skill. The execution flow includes at least one step that needs to be executed using an atomic skill. The atomic skill encapsulates executable logic code for implementing a predetermined function. When the atomic skill is used, the executable logic code is called to implement the predetermined function. The first execution module 44 is used to execute the steps using the atomic skills according to the arranged execution flow, so as to obtain the processing result of the target task.
[0074] In some embodiments, the thought chain skill includes a thought chain description, which includes at least one of a checklist, a sequence of steps, branching conditions, and feedback loop information, wherein: the checklist describes the content that needs to be checked at the checkpoints in the solution or the input information that needs to be obtained in the solution; the sequence of steps describes the execution steps in the solution; the branching conditions describe the logical judgment conditions for different execution paths in the solution; and the feedback loop information describes the processing principles for predetermined events.
[0075] In some embodiments, the atomic skills include data retrieval skills, table statistical analysis skills, report generation skills, report generation skills, image generation skills, and video script storyboard generation skills.
[0076] In some embodiments, the first processing module 42 is specifically used to: divide the target task into at least two task stages based on the target task processing request; For each of the task phases, identify the mind chain skills that are appropriate for that task phase; The second processing module 43 is specifically used to use the thought chain skills adapted to the task stage to arrange the execution flow of the task stage, wherein the execution flow of the task stage includes at least one step that needs to be executed using atomic skills. The first execution module 44 is specifically used to: for each task stage, according to the arranged execution flow of the task stage, use the atomic skills to execute the steps to obtain the execution result of each task stage; Based on the execution results of each of the task stages, the execution result of the target task is obtained.
[0077] In some embodiments, the task phase includes a first task phase and a second task phase that are executed sequentially. The device further includes: The decision module is used to evaluate whether the execution result of the first task stage meets the predetermined quality conditions after obtaining the execution result of the first task stage; if the execution result of the first task stage does not meet the predetermined quality conditions, determine the reason why the execution result does not meet the predetermined quality conditions, and determine the subsequent execution path based on the reason to adjust the execution result of the first task stage. The first execution module is specifically used to execute the steps using the atomic skills according to the arranged execution flow of the second task stage if the execution result of the first task stage meets the predetermined quality conditions, so as to obtain the execution result of the second task stage.
[0078] In some embodiments, the subsequent execution path includes: After adjusting the execution flow of the first task phase in the arrangement, the adjusted execution flow of the first task phase is re-executed. After rearranging the execution flow of the first task phase, the rearranged execution flow of the first task phase is executed. After adjusting the input data of the execution flow of the first task stage of the orchestration, the execution flow of the first task stage of the orchestration is re-executed.
[0079] In some embodiments, the first processing module is further configured to provide the target task processing request to the thought chain skill agent, so that the thought chain agent determines the thought chain skill that is compatible with the target task based on the task processing request, and then uses the thought chain skill to orchestrate the execution flow of the target task with a solution described by the thought chain skill.
[0080] In some embodiments, the first execution module is specifically configured to receive the orchestrated execution flow provided by the mind chain skill agent, and provide the orchestrated execution flow to the atomic skill agent so that the atomic skill agent performs the steps using the atomic skills according to the orchestrated execution flow; or, the mind chain agent provides the orchestrated execution flow to the atomic skill agent so that the atomic skill agent performs the steps using the atomic skills according to the orchestrated execution flow.
[0081] In this embodiment, the executable code is centralized, providing a unified, stable and monitorable execution environment for the invocation of atomic skills. At the same time, it is also beneficial to set up unified security policies and resource access control.
[0082] In some embodiments, the target task is a content marketing task, which includes industry insight tasks, competitor research tasks, content planning tasks, creator matching tasks, advertising placement tasks, and marketing effect review tasks.
[0083] See Figure 5 This application also provides an electronic device 50, which is based on the same technical concept as the foregoing embodiments. The electronic device provided in this application embodiment, such as... Figure 5 As shown, the steps and flow of the method described in any of the embodiments in Embodiment 1 of this application can be implemented.
[0084] The aforementioned electronic device may include a processor 51 and a memory 52, wherein the memory 52 is used to store executable program code; the processor 51 runs a program corresponding to the executable program code by reading the executable program code stored in the memory 52, for executing any of the task processing methods described in the foregoing embodiments.
[0085] For details on the specific execution process of the above steps by the processor 51 and the steps further executed by the processor 51 by running executable program code, please refer to the description of Embodiment 1 of this application, which will not be repeated here.
[0086] The electronic device exists in various forms, including but not limited to: (1) Mobile communication devices: These devices are characterized by their mobile communication capabilities and are primarily designed to provide voice and data communication. These terminals include smartphones (such as iPhones), multimedia phones, feature phones, and low-end phones.
[0087] (2) Ultra-mobile personal computer devices: These devices fall under the category of personal computers, possessing computing and processing capabilities, and generally also have mobile internet access features. These terminals include PDAs, MIDs, and UMPCs, such as the iPad.
[0088] (3) Portable entertainment devices: These devices can display and play multimedia content. This category includes audio and video players (such as iPods), handheld game consoles, e-book readers, as well as smart toys and portable car navigation devices.
[0089] (4) Server: A device that provides computing services. The components of a server include a processor, hard disk, memory, system bus, etc. Servers are similar to general computer architectures, but because they need to provide highly reliable services, they have higher requirements in terms of processing power, stability, reliability, security, scalability, and manageability.
[0090] (5) Other electronic devices with data interaction functions.
[0091] This application also provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the task processing method described in any of the preceding embodiments.
[0092] In summary, the embodiments of this application provide a task processing method, apparatus, electronic device, and storage medium. Skills are divided into atomic skills and thought chain skills. Solutions are encapsulated through thought chain skills, while executable logic code for various capabilities is encapsulated through atomic skills. During task processing, thought chain skills plan the task and, based on the encapsulated solution, orchestrate the task execution flow. Atomic skills are responsible for executing this flow. The solution is specialized, and the task processing process is standardized, thus effectively improving the reliability of task processing results. Furthermore, thanks to thought chain skills and atomic skills, human intervention is reduced in the task processing process, improving the automation level and processing efficiency.
[0093] Furthermore, encapsulating one or more executable logic codes into semantically clear atomic skills with explicit inputs and outputs enables upper-level agents to describe requirements in an abstract way without directly manipulating specific commands and files, significantly reducing system complexity and improving reusability.
[0094] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0095] The various embodiments in this specification are described in a related manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
[0096] For ease of description, if systems, servers, etc. are involved, they may be described separately as various units / modules based on their functions. Of course, in implementing this application, the functions of each unit / module can be implemented in one or more software and / or hardware.
[0097] 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 program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.
[0098] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A task processing method, characterized in that, The method includes: Receive the target task processing request input by the user; Based on the target task processing request, a thought chain skill that is compatible with the target task is determined, and the thought chain skill describes the solution to the target task; The execution flow of the target task is orchestrated using the thought chain skill to describe a solution based on the thought chain skill. The execution flow includes at least one step that requires the use of an atomic skill, wherein the atomic skill encapsulates executable logic code for implementing a predetermined function. When the atomic skill is used, the executable logic code is invoked to implement the predetermined function. According to the programmed execution flow, the atomic skills are used to perform the steps to obtain the processing result of the target task.
2. The task processing method according to claim 1, characterized in that, The thought chain skill includes a thought chain description, which includes at least one of a checklist, a sequence of steps, branching conditions, and feedback loop information, wherein: The checklist describes the contents that need to be checked at the checkpoints in the solution or the input information that needs to be obtained in the solution; The sequence of steps describes the execution steps in the solution; The branch conditions describe the logical judgment conditions for different execution paths in the solution; The feedback loop information describes the processing principles for predetermined events.
3. The task processing method according to claim 1, characterized in that, The atomic skills include data retrieval skills, table statistical analysis skills, report generation skills, report generation skills, image generation skills, and video script storyboard generation skills.
4. The task processing method according to claim 1, characterized in that, The process of determining the thought chain skills suitable for the target task based on the target task processing request includes: Based on the target task processing request, the target task is divided into at least two task stages; For each of the task phases, identify the mind chain skills that are appropriate for that task phase; The process of orchestrating the execution flow of the target task using the mind chain skills to describe the solution includes: Using the thought chain skills adapted to the task phase, the execution flow of the task phase is arranged, and the execution flow of the task phase includes at least one step that requires the use of atomic skills; The step of executing the steps according to the arranged execution flow, using the atomic skills, to obtain the processing result of the target task includes: For each of the task stages, the steps are executed using the atomic skills according to the arranged execution flow of the task stages to obtain the execution result of each task stage; Based on the execution results of each of the task stages, the execution result of the target task is obtained.
5. The task processing method according to claim 4, characterized in that, The task phase includes a first task phase and a second task phase, which are executed sequentially. After obtaining the execution result of the first task phase, the method further includes: Evaluate whether the execution results of the first task phase meet the predetermined quality conditions; If the execution result of the first task phase meets the predetermined quality conditions, the steps are executed using the atomic skills according to the arranged execution flow of the second task phase to obtain the execution result of the second task phase; If the execution result of the first task phase does not meet the predetermined quality conditions, determine the reason why the execution result does not meet the predetermined quality conditions, and determine the subsequent execution path based on the reason to adjust the execution result of the first task phase.
6. The task processing method according to claim 1, characterized in that, The process of determining the thought chain skills suitable for the target task based on the target task processing request includes: The target task processing request is provided to the thought chain skill agent, so that the thought chain agent determines the thought chain skill that is suitable for the target task based on the task processing request, and then uses the thought chain skill to arrange the execution flow of the target task with a solution based on the thought chain skill description.
7. The task processing method according to claim 6, characterized in that, The execution of the steps according to the arranged execution flow, using the atomic skills, includes: The system receives the orchestrated execution flow provided by the thought chain skill agent and provides the orchestrated execution flow to the atomic skill agent, so that the atomic skill agent executes the steps using the atomic skill according to the orchestrated execution flow. or, The thought chain agent provides the orchestrated execution flow to the atomic skill agent, so that the atomic skill agent can execute the steps using the atomic skills according to the orchestrated execution flow.
8. A task processing device, characterized in that, The device includes: The data receiving module is used to receive user-input requests for processing target tasks. The first processing module is used to determine, based on the target task processing request, a thought chain skill that is compatible with the target task, wherein the thought chain skill describes a solution to the target task; The second processing module is used to use the thought chain skill to arrange the execution flow of the target task based on the solution described by the thought chain skill. The execution flow includes at least one step that needs to be executed using an atomic skill. The atomic skill encapsulates executable logic code for implementing a predetermined function. When the atomic skill is used, the executable logic code is called to implement the predetermined function. The first execution module is used to execute the steps according to the arranged execution flow, using the atomic skills, to obtain the processing result of the target task.
9. An electronic device, characterized in that, The electronic device includes a processor and a memory, wherein the memory is used to store executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, for executing the task processing method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the task processing method according to any one of claims 1 to 7.