Task allocation method, device, equipment, medium and program product

Abstract

The embodiment of the invention relates to a task allocation method and device, equipment, a medium and a program product. The method includes generating a thinking chain related to a target task based on description of the target task and information of a plurality of task executors. The method further includes determining a task executor associated with the target task in the plurality of task executors through a task processing model based on the thinking chain. In addition, the method also includes assigning the target task to the determined task executor. According to the embodiment of the invention, the task is intelligently allocated through the task processing model, so that the task allocation efficiency can be improved, and the task execution quality can be improved. Moreover, the thinking chain is generated according to the description of the target task and the task executor information, so that the task is allocated by means of the task processing model, the task can be ensured to be allocated to a proper task executor, and the method is beneficial to reducing the cost of model training and improving the user experience.

Description

Technical Field

[0001] The present disclosure generally relates to the field of computers, and more particularly, to methods, devices, equipment, media, and program products for task allocation. Background Art

[0002] With the continuous progress of technology, machine learning models have become more complex and powerful. To fully utilize the performance of these models, more abundant and accurate labeled data needs to be provided, which has promoted the emergence and development of data annotation tasks. A data annotation task refers to the process of classifying, labeling, annotating, or commenting on data so that machine learning models can understand and use this data.

[0003] Data annotation plays a crucial role in multiple fields such as images, text, audio, and video. In the text field, data annotation may involve marking entities (such as person names, place names, organization names, etc.) in a sentence. In the image field, data annotation may include marking key information such as objects and scenes in an image. Data annotation tasks are usually completed by professional annotators or teams who use specialized annotation tools or platforms for annotation work. During the annotation process, it is necessary to ensure the accuracy, consistency, and integrity of the annotation to ensure that the trained machine learning model has high performance and reliability. Summary of the Invention

[0004] Embodiments of the present disclosure provide a method, device, equipment, medium, and program product for task allocation.

[0005] According to a first aspect of the disclosure, a method for task allocation is provided. The method includes generating a thought chain related to the target task based on the description of the target task and the information of multiple task executors. The method further includes determining, based on the thought chain, the task executor associated with the target task among the multiple task executors through a task processing model. In addition, the method further includes allocating the target task to the determined task executor.

[0006] In a second aspect of the disclosure, a device for task allocation is provided. The device includes a thought chain generation module configured to generate a thought chain related to the target task based on the description of the target task and the information of multiple task executors. The device further includes a task executor determination module configured to determine, based on the thought chain, the task executor associated with the target task among the multiple task executors through a task processing model. In addition, the device further includes a target task allocation module configured to allocate the target task to the determined task executor.

[0007] In a third aspect of the present disclosure, an electronic device is provided. The electronic device includes a processor and a memory coupled to the processor. The memory has instructions stored therein, and when the instructions are executed by the processor, the electronic device is caused to execute the method according to the first aspect.

[0008] In a fourth aspect of the present disclosure, a computer-readable storage medium is provided. Computer-executable instructions are stored on the computer-readable storage medium, and when the computer-executable instructions are executed by a processor, the method according to the first aspect is implemented.

[0009] In a fifth aspect of the present disclosure, a computer program product is provided. The computer program product is tangibly stored on a non-volatile computer-readable medium and includes computer-executable instructions, and when the computer-executable instructions are executed, a computer is caused to execute the method according to the first aspect.

[0010] The Summary section is intended to introduce a selection of concepts in a simplified form, which will be further described in the Detailed Description below. The Summary section is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to limit the scope of the claimed subject matter. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In conjunction with the accompanying drawings and with reference to the following detailed description, the above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent. In the drawings, the same or similar reference numerals denote the same or similar elements, where:

[0012] Figure 1 A schematic diagram showing an example environment in which multiple embodiments of the present disclosure can be implemented;

[0013] Figure 2 A flowchart showing a method for task allocation according to some embodiments of the present disclosure;

[0014] Figure 3 A schematic diagram showing an architecture for task allocation by means of an agent according to certain embodiments of the present disclosure;

[0015] Figure 4A A schematic diagram showing a chain of thought for task allocation according to certain embodiments of the present disclosure;

[0016] Figure 4B A schematic diagram showing a recommended allocation result and a recommended reason for allocation according to a chain of thought according to certain embodiments of the present disclosure;

[0017] Figure 5 A block diagram showing a device for task allocation according to certain embodiments of the present disclosure; and

[0018] Figure 6 A block diagram of an electronic device according to certain embodiments of the present disclosure is shown.

[0019] In all the figures, the same or similar reference numerals denote the same or similar elements. Detailed Description of Specific Embodiments

[0020] It can be understood that the data involved in the technical solution (including but not limited to the data itself, the acquisition or use of the data) should comply with the requirements of the corresponding laws, regulations and related provisions.

[0021] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although some embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not used to limit the protection scope of the present disclosure.

[0022] In the description of the embodiments of the present disclosure, the term "including" and its like should be understood as an open inclusion, that is, "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. may refer to different or the same objects, unless otherwise specified. There may also be other explicit and implicit definitions hereinafter.

[0023] In order to ensure that the machine learning model trained by the execution result obtained by executing the target task has high performance and reliability, the target task is usually completed by professional task executors or teams. In the related art, usually, after fusing the information features of all the executors who execute the target task with the information features of the target task to be processed, training is carried out in a related model, so that when applying, a task executor suitable for the target task can be found. The disadvantages of this method are obvious: First, considering that the number of executors who execute the target task is limited, and this method requires a large amount of training data, which may lead to overfitting, and these data require complicated feature transformation or encoding, resulting in the inability to improve the efficiency of task allocation. Second, considering the mobility of the executors who execute the target task, whenever there is a personnel change, the information features of the personnel will also change, which will cause the trained model to be invalid, and thus a new model needs to be trained again, wasting time and energy.

[0024] To this end, embodiments of the present disclosure provide a method for task allocation. In this method, a task processing model intelligently recommends a task executor for a target task based on a thought chain generated according to the description of the target task and the information of the task executor, so as to allocate the target task to the appropriate task executor.

[0025] In the embodiments of the present disclosure, by intelligently allocating tasks based on a thought chain through a task processing model, the efficiency of task allocation can be improved, making the work process more automated and efficient. Moreover, by matching according to the description of the target task and the information of the task executors (such as skills, specialties, etc.), it can be ensured that the tasks are allocated to suitable task executors. This personalized task allocation improves the quality and efficiency of task execution. In addition, when the task requirements or personnel capabilities change, the task allocation strategy can be quickly adjusted through the task processing model to adapt to the new situation.

[0026] Figure 1 FIG. shows a schematic diagram of an example environment 100 in which multiple embodiments of the present disclosure can be implemented. As Figure 1 shown, after the task allocation system 110 obtains the information 102 of multiple task executors and multiple task descriptions 104, it can recommend a relevant task executor for each task in parallel, and thus at 120, allocate the task to the corresponding task executor.

[0027] Referring to Figure 1 , in some embodiments, the task allocation system 110 may be an agent system, on which a task processing model is configured. An agent is an intelligent entity that can perceive the environment, make decisions, and execute actions. Usually based on machine learning technology, it has autonomy and adaptability and can autonomously learn and improve in specific tasks or domains. In some embodiments, the task processing model may be a machine learning base model trained with large-scale data. For example, it may be a language model or a multimodal model.

[0028] Continuing to refer to Figure 1 , in some embodiments, the information 102 of the task executor may be the skills of the task executor (such as language, expertise, etc.), past experience (such as experience in what types of tasks), historical task types, and accuracy rates (such as the mode of the accuracy rate in the image classification task is above 98%, while most of the other task executors are below 98%). In some embodiments, the task description 104 may be the standard operating procedure (SOP) of the task, the terms followed by the task (such as confidential information), the type of the task (such as annotation classification, rewriting, or extraction), the information of the task itself (such as video, text, picture, etc.), and unstructured descriptions (such as descriptions for videos or pictures), etc.

[0029] Figure 1 The task assignment system 110 shown in FIG. 1 can, according to the information 102 of task executors and the descriptions 104 of multiple tasks, recommend a relevant task executor for each task simultaneously. Thus, at 120, each task can be assigned to the corresponding task executor. The task assignment system 110 can efficiently match tasks and task executors to ensure that each task is properly handled.

[0030] It should be understood that the architecture and functions in the exemplary environment 100 are described only for exemplary purposes and do not imply any limitation on the scope of the present disclosure. Embodiments of the present disclosure can also be applied to other environments with different structures and / or functions.

[0031] The following will be combined with Figures 2 to 6 The method according to an embodiment of the present disclosure will be described in detail. For ease of understanding, the specific data mentioned in the following description are all exemplary and do not limit the protection scope of the present disclosure. It can be understood that the embodiments described below may also include additional actions not shown and / or actions shown may be omitted, and the scope of the present disclosure is not limited in this regard.

[0032] Figure 2 A flowchart of a method 200 for task assignment according to some embodiments of the present disclosure is shown. The method 200 can be executed by a device for task assignment, and the device can be implemented in software and / or hardware. Next, taking the device for task assignment as the execution subject as an example, the method 200 will be schematically described. Referring to Figure 2 , the method 200 may include block 202, block 204, and block 206.

[0033] In block 202, based on the description of the target task and the information of multiple task executors, a thought chain related to the target task is generated. For a target task, multiple task executors can be selected. In some embodiments, the target task can be a task to be assigned, such as a labeling task or a review task, and the task executors are the personnel responsible for executing these tasks, such as labelers or reviewers. The description of the target task covers various aspects of information closely related to the target task. In some embodiments, the description of the target task can be various descriptions related to the target task, such as the standard execution document of the target task, the content of the target task, or the type of the target task, etc. These task descriptions help to understand the essence and requirements of the task. In some embodiments, the information of the task executors can be the skills of the task executors, past experiences, or the types and accuracies of the tasks they have executed in the past, etc. In some embodiments, the thought chain at least includes the description of the target task and the information of the task executors. In some embodiments, the thought chain related to the target task refers to a series of logically related thinking steps or ideas related to the target task, which can guide the task processing model to determine the task executors related to the target task for the target task. In some embodiments, multiple thought chains associated with multiple target tasks can be determined in parallel.

[0034] In block 204, based on the thought chain, the task processing model determines the task executors associated with the target task among multiple task executors. That is to say, the task processing model selects the task executors related to the target task from among numerous task executors according to the thought chain. In some embodiments, the task processing model can be a machine learning-based model trained with large-scale data, such as a language model or a multimodal model. In some embodiments, multiple associated task executors can be determined in parallel for multiple target tasks. In some embodiments, the thought chain can have feedback information of error samples.

[0035] In block 206, the target task is assigned to the determined task executor. After determining the task executor associated with the target task for the target task, the target task can be assigned to this task executor. In some embodiments, multiple target tasks can be assigned to multiple task executors in parallel.

[0036] Embodiments of the present disclosure can improve the efficiency of task allocation and make the workflow more automated and efficient by using a task processing model to intelligently allocate tasks based on the chain of thought. Moreover, by matching the description of the target task with the information of the task executor (such as skills and specialties), it can ensure that the task is assigned to a suitable task executor, and this personalized task allocation can improve the quality and efficiency of task execution. In addition, when the task requirements or personnel capabilities change, the task processing model can quickly adjust the task allocation strategy to adapt to the new situation.

[0037] Figure 3 FIG. 300 shows a schematic diagram of an architecture 300 for task allocation by an agent according to some embodiments of the present disclosure. Hereinafter, taking the annotation task as the target task as an example, the process of task allocation by an agent in some embodiments of the present disclosure will be described. Therefore, in the description involving Figure 3 the task executor and the annotator can be referred to interchangeably.

[0038] Combined with Figure 1 , Figure 3 FIG. 110 shows a reference architecture of the task allocation system 110. The task allocation system 110 is configured with a task processing model 310, a memory 320, and a planning component 330. In some embodiments, the task processing model 310 can be a machine learning base model trained with a large amount of data, a multi-modal task processing model that can understand modal content data such as picture text, or a language model. In some embodiments, the task processing model 310 can be a model that generates text from input text, a model that generates text from input text and images, or a model that generates images from input text. In some embodiments, the task processing model 310 can be switched according to the type of the target task. For example, for a text classification task, a language model can be used.

[0039] Continuing to refer to Figure 3 FIG. 110, in some embodiments, the memory 320 is divided into a short-term memory 322 and a long-term memory 324. In some embodiments, the memory duration of the memory 320 can be set through the expiration duration of the database. For example, the expiration duration of the long-term memory is set to 1 month, and the expiration duration of the short-term memory is set to 1 day. By distinguishing between long-term memory and short-term memory, the storage cost can be saved, thereby improving the running efficiency of the task processing model 310.

[0040] As Figure 3 shown, in some embodiments, the original content of the task and the information of the task category can be stored in the short-term memory 322. In some embodiments, data such as the annotation results of the task executor (i.e., the annotator) and the verification results of quality assurance (QA) can also be stored in the short-term memory 322.

[0041] Continuing to refer toFigure 3 , in some embodiments, information such as the specialties and language skills of annotators is stored in the long-term memory 324, and statistical information on the performance of annotators in various tasks in historical annotation tasks can also be stored. For example, the average, median, mode, and variance of the accuracy rate of annotator 1 in short text classification tasks are 0.98, 0.94, 0.95, and 1.2 respectively. Another example is that the average, median, mode, and variance of annotator 1 in terms of annotation duration are 30 seconds, 40 seconds, 33 seconds, and 8.4 respectively. In some embodiments, the statistical information of these annotators can be updated periodically, for example, daily. In some embodiments, the long-term memory 324 can also store the standard operating procedure (SOP) of the annotation task, and the standard execution document can be updated as the annotation task type changes.

[0042] Continuing to refer to Figure 3 , in some embodiments, when preparing to execute an assigned annotation task, relevant information of the annotator such as skills can be pre-saved in the in-memory database of the memory 320. These databases can be NoSQL databases, such as in-memory data structure storage systems (such as Redis) or document-oriented NoSQL databases (such as MongoDB). In some embodiments, if there is also annotation information on the historical tasks of the annotator, it can also be saved in the database of the memory 320. In some embodiments, the relevant information about the annotator is stored in a discrete manner in the database of the memory 320. For example, "She is good at mathematics" can be stored in the database of the memory 320 as a discrete feature A. In some embodiments, when preparing to execute an assigned annotation task, the description related to the annotation task can also be stored in the memory 320. Some standard execution documents for the annotation task can be stored in the long-term memory 324, and some original content and task types of the annotation task can be stored in the short-term memory 322.

[0043] Such as Figure 3As shown, when an annotator needs to be assigned to annotation task B in annotation task 340, task assignment system 110 obtains various relevant descriptions of annotation task B and retrieves information on all annotators and historical annotation performance (if any) from the database in memory 320 to generate a chain of thought for this annotation task B. In some embodiments, discrete features of the information on annotators in memory 320 need to be converted into text expressions to participate in the construction of the chain of thought. For example, feature A can be converted into the text description "She is good at mathematics". In some embodiments, planning component 330 can be used to generate a chain of thought for annotation task B, and planning component 330 can also generate a chain of thought 334 for annotation task B according to some set templates. In some embodiments, some error feedback 332 on annotation tasks, after being verified by quality assurance, can be stored in memory 320 and can be retrieved by planning component 330 to generate a chain of thought 334 for annotation task B when relevant annotation tasks are assigned (such as when error feedback 332 is an annotation task of the same type as annotation task B).

[0044] Combined with Figure 4A , Figure 4A FIG. shows a schematic diagram of a chain of thought 400A for task assignment according to certain embodiments of the present disclosure. In display interface 402A, a chain of thought 334 for annotation task B is presented, which includes a Recommendation Instruction 410A, a task prompt 420A for annotation task B, a prompt 430A for all annotators, and a feedback prompt 440A. In some embodiments, exemplary feedback can also be added to the chain of thought 334, where exemplary feedback refers to tasks done well. Among them, the Recommendation Instruction 410A can have an instruction such as "Please output the ID of a suitable annotator". In some embodiments, the instruction can be changed according to the scenario of the target task.

[0045] Continuing to refer to Figure 4A , in some embodiments, the task prompt 420A can have a task description 422A of annotation task B. For example, the type is a text classification task, and the content of the task is the Portuguese text "Qual é a autonomia da bateria deste computador portátil". In some embodiments, the task prompt 420A can also include a standard operating procedure 424A for annotation task B, which instructs to classify the text into three categories according to rules, namely asking for product information, requesting help, and giving feedback.

[0046] Continuing to refer to Figure 4A, for annotation task B, the prompt 430A for the annotator may include the prompts of all annotators (from 1 to N). The prompt 432A of annotator 1 contains the relevant information of annotator 1: "She is very proficient in Portuguese. She was born in a Portuguese-speaking area. Her historical accuracy rate for short text classification is 98%, and the processing time is 23 seconds." The prompt 434A of annotator N contains the relevant information of annotator N: "He is good at Latin. Recently, his task accuracy rate for short text classification is 94%, and the processing time is 30 seconds." In some embodiments, the feedback prompt 440A is the error feedback 332 for some annotation tasks. For example, the feedback prompt 440A shows some historical error situations of annotator 3 for short text classification. When annotator 3 classified a Portuguese text, the text was wrongly marked as "Submit feedback", while the Quality Assurance (QA) marked it as "Request help". In some embodiments, the feedback prompt 440A can also be feedback such as an inappropriate annotator assigned to a certain task or a certain task being relatively difficult, resulting in an unsatisfactory annotation result. By adding feedback, a higher annotation accuracy rate and a shorter processing time can be obtained, thereby improving the overall efficiency of the annotation task. In some embodiments, if there is no error feedback 332 related to annotation task B, there may also be no error feedback prompt 440A in the chain of thought 334. In some embodiments, the parameters of the task processing model 310 can be adjusted according to the feedback of Quality Assurance.

[0047] By implementing the method of directly inputting the original data to assign annotators to the target task, it can ensure that the description of the input target task, the information of the annotators, the recommended results output, and the recommended reasons are controllable, thereby improving the efficiency of the system operation and enhancing the user experience.

[0048] The following will be combined with Figure 4B to illustrate the schematic diagram of the recommended assignment result and the recommended assignment reason 400B output according to the chain of thought in some embodiments of the present disclosure. Refer to Figure 4B , in the display interface 402B, the recommended reason 410B and the recommended result 420B are displayed. The recommended reason 410B output by the task processing model 310A can clearly show why annotator 1 is the suitable person for annotation task B rather than other annotators, improving the interpretability of the system. In some embodiments, if it is set not to output the recommended reason 410B, the recommended result 420B can also be directly output.

[0049] Return Figure 3, after assigning annotation task B to annotator 1, the annotation result 350 of annotator 1 for annotation task B can be sampled. For example, the annotation result 350 can be sampled at a certain ratio, such as 1%. After quality assurance, it can be determined whether the annotation of annotator 1 for annotation task B is accurate. In some embodiments, the judgment information of quality assurance will be stored in the memory data to form a short-term memory 322 for subsequent task assignment.

[0050] Continue to refer to Figure 3 , in some embodiments, if there are 500 annotators, the information of these 500 annotators can be divided into groups of 50 each and divided into 10 batches for the task recommendation system 110 to execute to select relevant annotators for annotation task B. Then, from the annotators determined in each batch as relevant to annotation task B, the final annotators suitable for annotation task B are determined. By this method of recommending suitable annotators for the target task, the controllability of the recommendation can be ensured, and compared with the method of recommending suitable annotation tasks for annotators, the overall reasoning efficiency is improved, thus bringing a better experience to users.

[0051] In some embodiments, there can be many target tasks in task 340. The task recommendation system 110 can parallelly determine which annotator matches these target tasks, and thus assign these target tasks to the corresponding annotators. It can be understood that the target task can also be other tasks, such as an audit task, etc.

[0052] By using a system configured with a task processing model 310 to recommend suitable annotators for the target annotation task, a large amount of sample data is not required, and it is not sensitive to changes in annotators. Thus, task recommendation and assignment can be achieved even when the annotators change.

[0053] Figure 5 shows a block diagram of a device 500 for task assignment according to certain embodiments of the present disclosure. As Figure 5 shown, the device 500 includes a thought chain generation module 502, configured to generate a thought chain related to the target task based on the description of the target task and the information of multiple task executors. The device 500 further includes a task executor determination module 504, configured to determine, based on the thought chain, the task executor among the multiple task executors associated with the target task through the task processing model. In addition, the device 500 further includes a target task assignment module 506, configured to assign the target task to the determined task executor.

[0054] In some embodiments, the chain of thought generation module 502 includes: an information acquisition module configured to acquire information of multiple task executors from the long-term memory of the memory; an error feedback acquisition module configured to acquire error feedback from the short-term memory of the memory, where the error feedback includes at least an error execution result and an error reason; and an agent chain of thought generation module configured to generate a chain of thought by the planning component of the agent based on the information, description, and error feedback.

[0055] In some embodiments, the task executor determination module 504 includes: a task processing model task executor determination module configured to determine, based on the chain of thought through a task processing model, a task executor associated with the target task.

[0056] In some embodiments, the task processing model task executor determination module includes: a reason determination module configured to, in response to the target task being a predetermined task, determine, based on the chain of thought through the task processing model, a task executor associated with the target task and a determination reason.

[0057] In some embodiments, the target task allocation module 506 includes: a labeled result acquisition module configured to acquire the labeled result of the target task by the determined task executor; and a feedback determination module configured to determine feedback for the labeled result based on the labeled result and store the feedback in the short-term memory.

[0058] In some embodiments, the feedback determination module includes: a sampling module configured to sample the labeled result according to a predetermined ratio; a feedback generation module configured to generate feedback based on the determination reason; and an adjustment module configured to adjust the parameters of the task processing model based on the feedback.

[0059] In some embodiments, the feedback determination module further includes an update module configured to update the error feedback to the planning component of the agent in response to the labeled result being inconsistent with the labeled result of quality assurance.

[0060] In some embodiments, the device 500 further includes: a data storage module configured to store the preprocessed data in the memory of the agent in a unified format, where the preprocessed data includes at least information of the task executor and a description.

[0061] In some embodiments, the device 500 further includes: a long-term memory determination module configured to determine that the memory is a long-term memory in response to the expiration duration of the database in the memory satisfying a first duration condition; and a short-term memory determination module configured to determine that the memory is a short-term memory in response to the expiration duration of the database satisfying a second duration condition.

[0062] In some embodiments, the chain of thought generation module 502 further includes: a discrete feature generation module configured to generate a plurality of discrete features for a plurality of task executors based on information of the plurality of task executors; and a discrete feature conversion module configured to convert the plurality of discrete features for the plurality of task executors into a plurality of text descriptions.

[0063] In some embodiments, the apparatus 500 further includes: a splitting module configured to split the plurality of task executors into a plurality of groups according to a unit quantity in response to the number of the plurality of task executors reaching a first predetermined quantity; a determining module configured to determine a plurality of candidate task executors batch by batch by an agent based on the split plurality of groups; and a second task executor determining module configured to determine a task executor associated with a target task from the plurality of candidate task executors.

[0064] Figure 6 FIG. shows a block diagram of an electronic device 600 according to certain embodiments of the present disclosure. The device 600 may be the device or apparatus described in the embodiments of the present disclosure. As Figure 6 shown, the device 600 includes a central processing unit (CPU) and / or a graphics processing unit (GPU) 601, which may execute various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) 602 or computer program instructions loaded from a storage unit 606 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device 600 may also be stored. The CPU / GPU 601, the ROM 602, and the RAM 603 are connected to each other through a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604. Although not shown in Figure 6 the device 600 may further include a coprocessor.

[0065] A plurality of components in the device 600 are connected to the I / O interface 605, including: an input unit 606, such as a keyboard, a mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a magnetic disk, an optical disc, etc.; and a communication unit 609, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0066] Each of the methods or processes described above may be executed by the CPU / GPU 601. For example, in some embodiments, the method may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed onto the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the CPU / GPU 601, one or more steps or actions of the methods or processes described above may be executed.

[0067] In some embodiments, the methods and processes described above may be implemented as a computer program product. The computer program product may include a computer-readable storage medium having thereon computer-readable program instructions for performing various aspects of the present disclosure.

[0068] A computer-readable storage medium may be a tangible device that can retain and store instructions for use by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disc (DVD), a memory stick, a floppy disk, a mechanically encoded device such as a punch card or raised structures in a groove having instructions stored thereon, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not construed to be a transitory signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through an optical fiber cable), or an electrical signal transmitted through a wire.

[0069] The computer-readable program instructions described herein may be downloaded to each computing / processing device from a computer-readable storage medium or may be downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network may include copper transmission cables, optical fiber transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in each computing / processing device.

[0070] The computer program instructions for performing the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine - related instructions, microcode, firmware instructions, state - setting data, or source code or object code written in any combination of one or more programming languages, including object - oriented programming languages and conventional procedural programming languages. The computer - readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a stand - alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network - including a local area network (LAN) or a wide area network (WAN) - or, alternatively, may be connected to an external computer (e.g., through the Internet using an Internet service provider). In some embodiments, by using the state information of the computer - readable program instructions to customize an electronic circuit, such as a programmable logic circuit, a field - programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit can execute the computer - readable program instructions to implement various aspects of this disclosure.

[0071] These computer - readable program instructions can be provided to a processing unit of a general - purpose computer, a special - purpose computer, or other programmable data - processing apparatus to produce a machine such that, when the instructions are executed by the processing unit of the computer or other programmable data - processing apparatus, a device is created that implements the functions / acts specified in one or more boxes of the flowchart and / or block diagram. The computer - readable program instructions can also be stored in a computer - readable storage medium, and these instructions cause a computer, a programmable data - processing apparatus, and / or other devices to work in a particular manner, so that the computer - readable medium storing the instructions comprises a manufacture, which includes instructions for implementing various aspects of the functions / acts specified in one or more boxes of the flowchart and / or block diagram.

[0072] The computer - readable program instructions can also be loaded onto a computer, other programmable data - processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data - processing apparatus, or other device to produce a computer - implemented process, whereby the instructions executed on the computer, other programmable data - processing apparatus, or other device implement the functions / acts specified in one or more boxes of the flowchart and / or block diagram.

[0073] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of devices, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a portion of an instruction, which contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions noted in the blocks may occur in a different order than noted in the accompanying drawings. For example, two consecutive blocks may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0074] The embodiments of the present disclosure have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, the practical application, or the technical improvement of the technology in the market, or to enable other ordinary skilled persons in the technical field to understand the embodiments disclosed herein.

Claims

1. A method for assigning tasks, comprising: Generate a thought chain related to the target task based on the description of the target task and the information of multiple task performers; Based on the thought chain, determining a task executor associated with the target task among the plurality of task executors through a task processing model; and The target task is assigned to the determined task executor.

2. The method according to claim 1, wherein generating a thought chain related to the target task based on the description of the target task and the information of multiple task performers comprises: Retrieving the information of the plurality of task performers from a long-term memory of a storage device; Acquire error feedback from the short-term memory of the memory, wherein the error feedback at least includes an error execution result and an error reason; as well as The thinking chain is generated by the planning component of the intelligent agent based on the information, the description and the error feedback.

3. The method according to claim 2, wherein the agent is configured with the task processing model, wherein based on the thought chain, determining the task executor associated with the target task among the plurality of task executors through the task processing model comprises: The task executor associated with the target task is determined based on the thought chain through the task processing model.

4. The method according to claim 3, wherein determining the task executor associated with the target task among the plurality of task executors through the task processing model comprises: In response to the target task being a predetermined task, the task processing model determines the task executor associated with the target task and the reason for determination based on the thought chain.

5. The method according to claim 4, wherein assigning the target task to the determined task performer comprises: Obtaining the marking result of the target task by the determined task executor; as well as Based on the annotation result, feedback for the annotation result is determined, and the feedback is stored in the short-term memory.

6. The method according to claim 5, wherein based on the annotation result, determining feedback for the annotation result comprises: Sampling the labeling results according to a predetermined ratio; generating the feedback based on the determination reason; as well as Based on the feedback, parameters of the task processing model are adjusted.

7. The method according to claim 6, further comprising: In response to the annotation result being inconsistent with the quality assurance annotation result, error feedback is updated to the planning component of the agent.

8. The method according to claim 1, further comprising: The preprocessed data is stored in a memory of the agent in a unified format, wherein the preprocessed data at least includes the information and the description of the task executor.

9. The method according to claim 8, further comprising: In response to the database expiration time of the memory satisfying a first time condition, determining that the memory is a long-term memory; as well as In response to the database expiration duration satisfying a second duration condition, it is determined that the memory is a short-term memory.

10. The method according to claim 8, wherein generating a thought chain related to the target task based on the description of the target task and the information of multiple task performers comprises: Based on the information of the multiple task performers, generating multiple discrete features for the multiple task performers; as well as The plurality of discrete features for the plurality of task performers are converted into a plurality of text descriptions.

11. The method according to claim 1, further comprising: In response to the number of the plurality of task executors reaching a first predetermined number, dividing the plurality of task executors into a plurality of groups according to the number of units; The intelligent agent determines a plurality of candidate task executors in batches based on the plurality of split groups; as well as The task performer associated with the target task is determined from among the plurality of candidate task performers.

12. A device for allocating tasks: A thought chain generation module is configured to generate a thought chain related to the target task based on a description of the target task and information of multiple task performers; A task executor determination module is configured to determine a task executor associated with the target task among the plurality of task executors through a task processing model based on the thought chain; as well as The target task allocation module is configured to allocate the target task to the determined task executor.

13. An electronic device comprising: processor; as well as A memory coupled to the processor, the memory having instructions stored therein, wherein when the instructions are executed by the processor, the electronic device executes the method according to any one of claims 1 to 11.

14. A computer-readable storage medium having computer-executable instructions stored thereon, wherein the computer-executable instructions are executed by a processor to implement the method according to any one of claims 1 to 11.

15. A computer program product tangibly stored on a non-transitory computer readable medium and comprising computer executable instructions which, when executed, cause a computer to perform the steps of the method according to any one of claims 1 to 11.

Images