A task processing method, device, system, electronic device and storage medium

Abstract

The present disclosure provides a task processing method, apparatus, system, electronic device, and storage medium, which relate to the field of computer technologies, specifically to the fields of distributed computing technologies, big data, and artificial intelligence technologies such as deep learning. Among them, the task processing method includes: obtaining task request information sent by a user terminal according to a task to be processed; generating a model interaction request in real time according to the task request information; sending the model interaction request to a model server in real time; receiving a current model processing result feedback by the model server in real time according to the model interaction request; and determining a current task processing result of the task to be processed according to the current model processing result. The embodiments of the present disclosure can improve the real-time performance and efficiency of task processing.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technologies, specifically to the fields of distributed computing, big data, and artificial intelligence technologies such as deep learning, and particularly to a task processing method, apparatus, system, electronic device, computer-readable storage medium, and computer program product. Background Art

[0002] With the development of science and technology, models are increasingly widely used. For example, in the field of artificial intelligence technologies, various network models can be developed and trained in a deep learning manner, and the developed and trained network models can be applied to various application scenarios such as image recognition, speech recognition, and target detection. For another example, in the field of data processing technologies, various calculation models can also be developed according to business requirements, so as to automatically calculate reasonable solutions based on the input configuration data through the calculation models, thereby providing automated and intelligent calculation services. In the case of an increasing number of model types, how to reasonably schedule and adapt the appropriate models and determine a reasonable interaction method with the models to provide high-quality model services has become a key issue of concern in the current model scheduling and interaction field. Summary of the Invention

[0003] Embodiments of the present disclosure provide a task processing method, apparatus, system, electronic device, and storage medium, which can improve the real-time performance and efficiency of task processing.

[0004] In a first aspect, embodiments of the present disclosure provide a task processing method, including:

[0005] Obtaining task request information sent by a client according to a task to be processed;

[0006] Generating a model interaction request in real time according to the task request information;

[0007] Sending the model interaction request to a model server in real time;

[0008] Receiving a current model processing result feedback by the model server in real time according to the model interaction request;

[0009] Determining a current task processing result of the task to be processed according to the current model processing result.

[0010] In a second aspect, embodiments of the present disclosure provide a task processing apparatus, including:

[0011] A task request information obtaining module, configured to obtain task request information sent by a client according to a task to be processed;

[0012] A model interaction request generating module, configured to generate a model interaction request in real time according to the task request information;

[0013] A model interaction request sending module, configured to send the model interaction request to a model server in real time;

[0014] A model current processing result receiving module, configured to receive the model current processing result that the model server feeds back in real time according to the model interaction request;

[0015] A current task processing result determining module, configured to determine the current task processing result of the to-be-processed task according to the model current processing result.

[0016] Thirdly, an embodiment of the present disclosure provides a task processing system, including a user terminal, a server, and a model server; the user terminal is communicatively connected to the server, and the server is communicatively connected to the model server; wherein:

[0017] The user terminal is configured to generate task request information according to a to-be-processed task, and send the task request information to the server;

[0018] The server is configured to generate a model interaction request in real time according to the task request information, and send the model interaction request to the model server in real time;

[0019] The model server is configured to feed back the model current processing result in real time according to the model interaction request, and feed back the model current processing result to the server;

[0020] The server is further configured to determine the current task processing result of the to-be-processed task according to the model current processing result.

[0021] Fourthly, an embodiment of the present disclosure provides an electronic device, including:

[0022] At least one processor; and

[0023] A memory communicatively connected to the at least one processor; wherein,

[0024] The memory stores instructions executable by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor is enabled to execute the task processing method provided in the embodiment of the first aspect.

[0025] Fifthly, an embodiment of the present disclosure further provides a non-transitory computer-readable storage medium storing computer instructions, where the computer instructions are used to cause a computer to execute the task processing method provided in the embodiment of the first aspect.

[0026] In a sixth aspect, an embodiment of the present disclosure further provides a computer program product, including a computer program, which when executed by a processor, implements the task processing method provided in the embodiment of the first aspect.

[0027] After obtaining the task request information sent by the client according to the task to be processed, the embodiment of the present disclosure generates a model interaction request in real time according to the obtained task request information, and sends the real-time generated model interaction request to the model server in real time, so that the model server processes the task to be processed according to the model interaction request, and receives the current processing result of the model feedback by the model server in real time according to the model interaction request, thereby determining the current task processing result of the task to be processed according to the received current processing result of the model, solving the problems such as low real-time performance and efficiency of task processing when processing tasks according to the model, and being able to improve the real-time performance and efficiency of task processing.

[0028] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The drawings are used to better understand the solution and do not constitute a limitation to the present disclosure. Among them:

[0030] Figure 1 is a schematic flowchart of a process for processing tasks through a synchronous waiting interaction model call mode in the related art;

[0031] Figure 2 is a schematic flowchart of a process for calling a model to process tasks through a front-end asynchronous polling mode in the related art;

[0032] Figure 3 is a flowchart of a task processing method provided by an embodiment of the present disclosure;

[0033] Figure 4 is a flowchart of a task processing method provided by an embodiment of the present disclosure;

[0034] Figure 5 is a flowchart of a task processing method provided by an embodiment of the present disclosure;

[0035] Figure 6 is a flowchart of a task processing method provided by an embodiment of the present disclosure;

[0036] Figure 7 is a flowchart of a task processing method provided by an embodiment of the present disclosure;

[0037] Figure 8 is a structural diagram of a task processing device provided by an embodiment of the present disclosure;

[0038] Figure 9 It is a structural diagram of a task processing system provided by an embodiment of the present disclosure;

[0039] Figure 10 It is a schematic diagram of the architecture of a task processing system provided by an embodiment of the present disclosure;

[0040] Figure 11 It is a schematic diagram of the effect of a task processing interface provided by an embodiment of the present disclosure;

[0041] Figure 12 It is a schematic diagram of the architecture for distributed scheduling of a server-side adaptation model on the server side provided by an embodiment of the present disclosure;

[0042] Figure 13 It is a schematic diagram of the process of a worker executing a task in an Airflow scheduling model provided by an embodiment of the present disclosure;

[0043] Figure 14 It is a schematic diagram of the process of Spark distributed computing provided by an embodiment of the present disclosure;

[0044] Figure 15 It is a schematic diagram of the structure of an electronic device for implementing the task processing method of an embodiment of the present disclosure. Detailed implementation manners

[0045] The following makes an explanation of the exemplary embodiments of the present disclosure with reference to the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted below.

[0046] When a user needs to process certain tasks according to their own demand information, they often need to submit the demand information to the model server to call a matching model to automatically process the to-be-processed tasks delivered by the user according to the user's demand information. For example, the user submits an image to be recognized to the model server to call a matching image recognition model to automatically perform image recognition or target detection on the image to be recognized. Another example is that the user submits the configuration or attribute information of a solution (such as a bidding solution, a data mining and analysis solution, or an investment project solution, etc.) to the model server to call a matching image recognition model to automatically formulate a reasonable solution for the user based on the information submitted by the user for the user to refer to, understand, configure, or optimize the solution calculated by the model.

[0047] Currently, when calling a model to calculate the to-be-processed tasks delivered by a user, the commonly used model interaction method is the synchronous waiting interaction mode, Figure 1It is a schematic flow chart of the related technology for processing tasks through the synchronous waiting interaction model call mode. In a specific example, such as Figure 1 shown, the general process of the synchronous waiting interaction mode is as follows: The user configures the relevant information of the task to be processed at the front end of the platform and initiates a model calculation request. The back end of the platform verifies the information configured by the user, and after passing the verification, calls the model-side interface to call the matching model for the calculation task. After receiving the calculation request, the model server obtains the supporting data according to the relevant information of the task to be processed. For example, when the task to be processed is an investment project, the market data and product data in the project field can be obtained in real time, and the matching model calculation result is scheduled according to the relevant configuration information of the task to be processed and the obtained supporting data. Or, the model server can also directly schedule the matching model calculation result according to the relevant configuration information of the task to be processed. The front end of the platform is always in a waiting state during the model calculation process. During this period, any operation by the user on the front end of the platform will cause the calculation to terminate. It is necessary to wait until the model-side calculation is completed, and the front end of the platform obtains the result return, renders the interface, for the user to view and analyze, and continue the operation.

[0048] The above task processing method of the synchronous waiting interaction mode has the following significant problems: (1) It is necessary to maintain long connections at the front end, the server end, and the model server end. When the data volume of the task to be processed is too large and the calculation time interval is too long, the user needs to wait for a long time for the return of the calculation result and cannot perform any operations during this period, seriously affecting the timeliness of task processing and thus affecting the user experience. (2) For a long time calculation, the connection and thread pressure on the front end, the back end, and the model end are too large, the system fault tolerance is poor, and the probability of being affected by factors such as network jitter is relatively large, and the failure rate is high. (3) The logic of synchronous calculation cannot fully utilize the server resources. For scenarios involving big data calculations, the advantages of distributed calculation cannot be exerted, resulting in poor scalability. When the task pressure is too large, the interface-based model service cannot effectively schedule the model to execute tasks, and the excessive calculation pressure may cause the server to crash.

[0049] To solve the above problems, the related technology proposes a front-end asynchronous polling mode to call the model to process the task to be processed submitted by the user. Figure 2 It is a schematic flow chart of the related technology for calling the model to process tasks through the front-end asynchronous polling mode. In a specific example, such as Figure 2As shown, after the user configures the information of the task to be processed at the front end of the platform and submits a model calculation request, a server layer is added between the front end of the platform and the model server. The server maintains a long connection with the interface of the model server through an asynchronous thread and waits for the calculation results to be returned from the model side. At the same time, the front end polls the synchronous results of the server in a timed polling manner. When the model server finishes the calculation, the server updates the calculation results to the database for the user to request and view. This method supports the user to perform other operations on the system after submitting the model calculation and can return at any time to view the model calculation results.

[0050] However, the task processing method of the above front-end asynchronous polling mode still has the following significant problems: (1) Poor service stability. The front-end asynchronous polling mode also needs to maintain a long connection between the server and the model server, resulting in large thread overhead and connection overhead, high pressure on the model-side interface, and the service is prone to crash under high pressure. (2) Low computing efficiency. The front-end asynchronous polling mode only optimizes the user experience to a certain extent, but does not fundamentally solve the problems of slow model calculation efficiency, high calculation pressure, and uneven distribution of computing resources.

[0051] After debugging the models with slow execution speed, model developers found that the reason for the slow execution of some models lies in the reading and writing speed of the underlying data. Therefore, in order to improve the calculation efficiency of the models, related technologies have also proposed an optimized storage mode to call the models to process the tasks to be processed submitted by the users. Specifically, indexes can be added to the fields with high query frequencies, thereby effectively improving the execution efficiency of some parts. For data combinations with sparse data, columnar storage such as HBase (a distributed, column-oriented open-source database) or Hive (a data warehouse tool) is used to optimize the underlying data storage configuration, effectively reducing the read and write I / O and improving the read and write efficiency. At the same time, the scale of the calculated data is controlled, so the calculation efficiency is improved to a certain extent.

[0052] However, optimizing the storage mode requires migrating the underlying data. In some environments, databases such as Hbase or Hive need to be built. The storage interface layer of the algorithm needs to be modified, and the calculation logic also needs to be adapted and modified accordingly, resulting in a high transformation cost.

[0053] Considering that the calculation speed of most models is slow, the core reason is that their computational volume is large and the calculation logic is complex. When the calculation logic cannot be optimized, model developers consider optimizing the calculation efficiency of the model by optimizing the hardware configuration. A midrange computer uses a reduced instruction set processor and is a high-performance 64-bit computer that supports single-point high-performance business calculations and is suitable for highly reliable industry applications such as financial securities. After optimizing the hardware configuration with a midrange computer, the calculation efficiency of the model has been greatly improved, and it can basically cover the calculation scenarios of most models.

[0054] However, the method of optimizing the midrange computer configuration requires the independent purchase of a midrange computer for the model's calculation environment, resulting in increased hardware costs. At the same time, since the single-point calculation mode is adopted, the advantages of distributed calculation cannot be fully utilized, and the calculation efficiency of the model cannot be improved from the software level.

[0055] In one example, Figure 3 is a flowchart of a task processing method provided by an embodiment of the present disclosure. This embodiment is applicable to the situation where a model is called in a real-time interaction manner to process a to-be-processed task of a user-side interaction. This method can be executed by a task processing device, which can be implemented in a software and / or hardware manner and is generally integrated in an electronic device. The electronic device can be a server device and is used in cooperation with a user side and a model server side. Correspondingly, as Figure 3 shown, the method includes the following operations:

[0056] S310, Obtain task request information sent by the user side according to the to-be-processed task.

[0057] Among them, the user side is the front end of the platform facing the user, which can interface with the user to receive relevant information for processing tasks sent by the user. The to-be-processed task can be a task that the user submits on the user side and needs to call the model for processing. The task type of the to-be-processed task can be, for example, an image recognition task, a data calculation, or a project planning task, etc. The embodiment of the present disclosure does not limit the task type of the to-be-processed task. The task request information can be the request information for the user to request the model to process the to-be-processed task for the to-be-processed task.

[0058] When a user needs to process certain tasks according to their own demand information, they can operate on the user side and submit relevant information of the to-be-processed task based on the user side. After the user side receives the to-be-processed task submitted by the user, it can generate matching task request information according to the to-be-processed task and send it to the server to request the server to schedule a matching model to process the to-be-processed task according to the task request information.

[0059] Exemplarily, the user submits a data calculation task at the user side, and provides the data source required for the data calculation task and the output format information of the calculation result. Then, the user side can generate corresponding data calculation request information based on the data source and the output format information of the calculation result matched by the data calculation task, and send it to the server side, so as to request the server side to schedule the matched mathematical calculation model to process the data calculation task and sort out the calculation result of the data calculation task according to the output format information required by the user, and obtain the final task processing result of the data calculation task.

[0060] In the embodiments of the present disclosure, the task to be processed may be a task newly created by the user that needs to start processing, or a task for the user to query the status of the task being processed by the model, or a task for the user to delete the task being processed by the model. The embodiments of the present disclosure do not limit this. That is, for the tasks that have been submitted and are being processed, the user does not need to wait for the execution results of the tasks and can process other tasks, and can synchronously and parallelly execute multiple different tasks, thereby improving the task processing efficiency.

[0061] S320. Generate a model interaction request in real time according to the task request information.

[0062] Among them, the model interaction request may be a request for the server side to schedule the model to process the task to be processed.

[0063] S330. Send the model interaction request to the model server in real time.

[0064] Among them, the model server may be the backend that provides model services. The model server may integrate multiple models in different fields and of different types. For example, it may integrate multiple image recognition models in the field of image recognition, and may also integrate multiple data calculation models in the field of data calculation, as long as they are models that can be scheduled and applied. The embodiments of the present disclosure do not limit the types and quantities of the models integrated by the model server.

[0065] Correspondingly, after the server side receives the task request information sent by the user side according to the task to be processed, it can generate a model interaction request in real time according to the task request information, and send the generated model interaction request to the model server in real time, so as to schedule the model matched by the model server to process the task to be processed.

[0066] Exemplarily, assume that the user submits a to-be-processed task for image recognition. After the server receives the task request information corresponding to image recognition, it can generate a model interaction request for scheduling the image recognition model in real time, and send the model interaction request to the model server, so as to call the interface between the server and the model server through the model interaction request, schedule the image recognition model that best matches the task request information corresponding to image recognition through the called interface, and submit the specific processing information of the to-be-processed task, such as the image to be recognized, etc., to the scheduled image recognition model through the model interaction request.

[0067] S340. Receive the current model processing result that the model server feeds back in real time according to the model interaction request.

[0068] Among them, the current model processing result may be the task processing result that the model server generates in real time for the to-be-processed task according to the model interaction request.

[0069] Correspondingly, after the model server receives the model interaction request, it can start processing the to-be-processed task. To achieve real-time response to the to-be-processed task, regardless of whether the to-be-processed task is executed to completion to obtain an execution result, the model server needs to obtain the current processing result of the to-be-processed task in real time as the current model processing result, and feed back the current model processing result to the server in real time.

[0070] Exemplarily, assume that the to-be-processed task has not been processed to completion. Then the model server can generate the result of "task in execution" in real time as the current model processing result and feed it back to the server in real time. Assume that the to-be-processed task has been processed to completion. Then the model server can generate the specific result data of the to-be-processed task in real time as the current model processing result and feed it back to the server in real time.

[0071] S350. Determine the current task processing result of the to-be-processed task according to the current model processing result.

[0072] Among them, the current task processing result may be the task processing result generated by the server according to the current model processing result that the model server feeds back in real time.

[0073] Correspondingly, after the server receives the current model processing result that the model server feeds back in real time according to the model interaction request, it can further process the current model processing result to determine the current task processing result of the to-be-processed task. It can be understood that according to the different types and execution states of the to-be-processed tasks, the current task processing result generated by the server can be stored locally or fed back to the user side in real time. The embodiments of the present disclosure do not limit this.

[0074] Exemplarily, assuming that the current processing result of the model is "Task in progress", the server can generate the current task processing result of "The task has not been completed yet, please wait" and feedback it to the client. Assuming that the current processing result of the model is the specific result data of the task to be processed, the server can further organize the specific result data of the task to be processed to obtain the final result data. For example, the specific result data of the task to be processed is further organized according to the data output format specified by the user to obtain the current task processing result, and the previous task processing result is feedback to the client.

[0075] In the embodiments of the present disclosure, in order to avoid the failure of the task to be processed, during the process of processing the task to be processed, there is no need to maintain a long connection between the client and the server, and between the server and the model server. Instead, a real-time request and real-time response method is used for data interaction. That is, when the user submits a task to be processed through the client, the connection between the client and the server is disconnected. The user does not need to wait for the processing result of the current task and can still submit other new tasks to be processed on the client, or query the status of the task being processed by the model and delete the task, etc. for subsequent executable tasks of the task to be processed, thereby improving the real-time performance and efficiency of task processing, and further improving the user experience. After the server sends a model interaction request to the model server, the model server can independently process the task to be processed in response to the model interaction request, and regardless of whether the task to be processed is completed, it can respond to the model interaction request in real time and feedback the current processing result of the model to the server. When the server receives the current processing result of the model, the connection between the server and the model server is disconnected. At this time, the model server can still continue to process the task to be processed in the case of disconnection.

[0076] It can be seen that the task processing method provided by the embodiments of the present disclosure belongs to a completely asynchronous system interaction mode, which reduces the coupling between the client, the server, and the model server, and the module functions are clearer. The long connection data interaction method between each module is abandoned, which can not only improve the real-time performance and efficiency of task processing, but also effectively ensure the stability of the task processing service.

[0077] After obtaining the task request information sent by the client according to the task to be processed in the embodiments of the present disclosure, a model interaction request is generated in real time according to the obtained task request information, and the real-time generated model interaction request is sent to the model server in real time, so that the model server processes the task to be processed according to the model interaction request, and receives the current processing result of the model feedback by the model server in real time according to the model interaction request, thereby determining the current task processing result of the task to be processed according to the received current processing result of the model, and solving the problems such as low real-time performance and efficiency of task processing in the related art according to the model, and being able to improve the real-time performance and efficiency of task processing.

[0078] In one example, Figure 4 is a flowchart of a task processing method provided by an embodiment of the present disclosure. Based on the technical solutions of the above embodiments, the embodiment of the present disclosure has been optimized and improved. When the task to be processed is a new task, multiple specific and optional implementation manners are given for obtaining task request information sent by the client according to the task to be processed, generating a model interaction request in real time according to the task request information, receiving the current processing result of the model feedback by the model server in real time according to the model interaction request, and determining the current task processing result of the task to be processed according to the current processing result of the model.

[0079] Such as Figure 4 shown, a task processing method includes:

[0080] S410. Obtain task request information sent by the client according to the new task.

[0081] Among them, the new task may be a to-be-processed task newly initiated by the user on the client.

[0082] It can be understood that the user can submit a new task on the client as the task to be processed. At this time, the server can receive the task request information sent by the client according to the new task.

[0083] S420. Generate a model interaction request in real time according to the task request information.

[0084] Correspondingly, step S420 may specifically include the following operations:

[0085] S421. Obtain the task configuration information of the new task according to the task request information.

[0086] Among them, the task configuration information may be relevant configuration information submitted by the user through the client for the new task.

[0087] Correspondingly, when the server receives the new task, it can obtain the task configuration information of the new task according to the task request information of the new task.

[0088] Exemplarily, when the user takes the data calculation task as the new task, the task configuration information of the new task may be information such as the data source to be calculated and the specific output format of the calculation result. When the user takes the project planning task of the investment project as the new task, the task configuration information of the new task may be asset configuration information such as the investment portfolio list, benchmark type, currency type, penetration type, and date.

[0089] S422. Determine the first target model for processing the new task according to the task configuration information.

[0090] Among them, the first target model can be a model called by the server from the model server for processing a new task submitted by a user.

[0091] S423. Generate the model scheduling request in real time according to the first target model.

[0092] After the server obtains a new task and the task configuration information of the new task, it can determine the first target model in the model server for processing the new task according to the type of the new task and the specific task configuration information according to a preset model scheduling strategy. Exemplarily, the server can determine multiple alternative models that can be used to process the new task according to the type of the new task, and further screen the most matching model from each alternative model as the first target model according to the task configuration information. Correspondingly, after determining the first target model, the server can generate a model scheduling request in real time according to the determined first target model to schedule the first target model to process the new task through the model scheduling request.

[0093] In the embodiment of the present disclosure, after the server generates a model interaction request in real time, it can disconnect the connection with the user side for the new task event. The server can independently execute the subsequent model scheduling process according to the received information, and the user side does not need to wait for the server to feedback the processing result of the new task and can synchronously process other types of pending tasks. The advantage of such a setting is that it can avoid the problem of task processing failure caused by network jitter delay.

[0094] Through the above technical solution, by generating a model scheduling request for processing a new task in real time for a new task submitted by a user, and scheduling the first target model to process the new task through the model scheduling request, the timeliness of new task processing can be ensured.

[0095] S430. Send the model scheduling request to the model server in real time.

[0096] S440. Receive the task identifier that the model server feedbacks in real time according to the model scheduling request.

[0097] Among them, the task identifier can be an identifier generated by the model server for uniquely identifying a pending task. This identifier can be in the form of a task number, as long as it can uniquely identify the new task. The embodiment of the present disclosure does not limit the identifier type and content of the task identifier.

[0098] In the embodiment of the present disclosure, after the model server receives the model scheduling request, it can obtain the relevant information of the new task according to the model scheduling request to generate a task identifier matching the new task, and feedback the task identifier to the server in real time.

[0099] Correspondingly, after the model server feeds back the task identifier to the server in real time, the connection with the server can be disconnected, and the first target model can be independently used to execute the new task without connection to the server. The advantage of this setting is that it can avoid the problem of task processing failure caused by service crashes due to large thread overhead and connection overhead.

[0100] S450. Determine the current task processing result of the to-be-processed task according to the current processing result of the model.

[0101] Correspondingly, step S450 may specifically include the following operations:

[0102] S451. Obtain the task query identifier sent by the user terminal according to the new task.

[0103] Among them, the task query identifier may be the identifier information used by the user terminal to query the execution result of the to-be-processed task.

[0104] If the to-be-processed task is a new task, the user terminal may generate a matching task query identifier for each new task to query the subsequent execution result of the new task. After the user terminal generates the task query identifier, it may send the task query identifier to the server.

[0105] In an optional embodiment of the present disclosure, the obtaining the task query identifier sent by the user terminal according to the to-be-processed task may include: obtaining the space factor identifier, user identifier, and model identifier sent by the user terminal according to the to-be-processed task; generating the task query identifier according to the space factor identifier, the user identifier, and the model identifier.

[0106] Among them, the space factor identifier may be an identifier type for annotating the to-be-processed task from the perspective of space factors. The user identifier may be an identifier type for annotating the to-be-processed task from the perspective of the user. The model identifier may be an identifier type for annotating the to-be-processed task from the perspective of the model.

[0107] Specifically, the client can determine a space factor identifier (abbreviated as spaceID), a user identifier (abbreviated as userID), and a model identifier (abbreviated as moduleID) for the task to be processed. For example, the client generates a space factor identifier based on the space occupied by the task to be processed, generates a user identifier based on the user information initiating the task to be processed, and generates a model identifier based on the model information (such as the type of the model, etc.) required for the task to be processed. Further, the client can send the space factor identifier, the user identifier, and the model identifier to the server, and the server generates a task query identifier matching the new task according to the space factor identifier, the user identifier, and the model identifier. Alternatively, the client can also generate a task query identifier matching the new task according to the space factor identifier, the user identifier, and the model identifier. When the server generates the task query identifier, the server needs to feedback the generated task query identifier to the client.

[0108] Optionally, generating a task query identifier matching the new task according to the space factor identifier, the user identifier, and the model identifier can be any form of permutation and combination of the space factor identifier, the user identifier, and the model identifier, or a combination of the space factor identifier, the user identifier, and the model identifier based on other associated information used to generate the task query identifier. The embodiments of the present disclosure do not limit the manner of generating the task query identifier.

[0109] Through the above technical solution, by generating a task query identifier according to the space factor identifier, the user identifier, and the model identifier, it is convenient for the client to query the execution status of the task in real time later and improve the task response efficiency.

[0110] S452. Establish and store an identification mapping relationship between the task query identifier and the task identifier.

[0111] Among them, the identification mapping relationship can be a mapping relationship between the task query identifier and the task identifier.

[0112] S453. Generate the current task processing result according to the task query identifier, the task identifier, and the identification mapping relationship.

[0113] It can be understood that, under normal circumstances, the first target model requires a certain amount of time to process a newly created task. Especially when the newly created task and / or the first target model are relatively complex, it often requires a long task processing time. Therefore, the server may not be able to feedback the final processing result of the newly created task to the user side in real time. Correspondingly, after the server obtains the task query identifier of the newly created task and the task identifier feedback by the model server according to the newly created task, it can establish and store the identifier mapping relationship between the task query identifier and the task identifier, and generate the current task processing result according to the task query identifier, the task identifier, and the identifier mapping relationship. Since the newly created task has not been completed yet, the server can directly cache the current task processing result locally. At this time, since there is no connection for the newly created task established between the user side and the server, the user side can display the loading state of the newly created task on the task processing interface of the newly created task to wait for the model server to complete the calculation.

[0114] In the above technical solution, after the user side submits a newly created task, it disconnects from the server. The server generates a model scheduling request for scheduling the first target model according to the task configuration information of the newly created task, and schedules the first target model of the model server to independently process the newly created task without a connection through the model scheduling request, which can avoid the problem of task processing failure caused by network jitter delay, and the problem of task processing failure caused by service crash due to large thread overhead and connection overhead, thereby ensuring the success rate of task execution and improving the real-time performance and efficiency of task processing.

[0115] In one example, Figure 5 is a flowchart of a task processing method provided by an embodiment of the present disclosure. Based on the technical solutions of the above embodiments, the embodiment of the present disclosure is optimized and improved. In the case where the task to be processed is an execution status query task, multiple specific and optional implementation manners of obtaining the task request information sent by the user side according to the task to be processed, generating a model interaction request in real time according to the task request information, receiving the current model processing result feedback by the model server in real time according to the model interaction request, and determining the current task processing result of the task to be processed according to the current model processing result are given.

[0116] Such as Figure 5 shown, a task processing method includes:

[0117] S510. Obtain the task request information sent by the user side according to the execution status query task.

[0118] Among them, the execution status query task may be a task for querying the status of the task being processed by the model.

[0119] In an embodiment of the present disclosure, in addition to submitting a new task, the client can also query the status of a task being processed by the model, thereby submitting an execution status query task. Since the connection between the client and the server is disconnected after the client submits a task, the client needs to initiate an execution status query task later to obtain the execution result of the task. Correspondingly, the server can receive the task request information sent by the client according to the execution status query task. After the client submits the task request information of the execution status query task, the connection with the server can be disconnected.

[0120] In an optional embodiment of the present disclosure, the obtaining of the task request information sent by the client according to the task to be processed may include: obtaining the task request information of the execution status query task sent by the client at a set polling period; and / or obtaining the task request information of the execution status query task sent by the client in response to a user query operation.

[0121] Among them, the set polling period may be a polling query period set by the client for the task being processed.

[0122] When the user submits a new task on the client, if the user always stays on the task processing interface of the new task without performing other task processing operations, or when the user stays on other task processing interfaces waiting for the task execution result, in order to avoid being unable to obtain the processing result of the new task in real time due to the disconnection of the connection, the client can send the task request information of the execution status query task for querying the task execution status to the server at a set polling period, so as to periodically query and obtain the task execution status of the current processing task corresponding to the current task processing interface.

[0123] Exemplarily, the client can poll and query the execution status of the task to be queried by the execution status query task from the server every 5 minutes. Among them, the set polling period can be set according to actual business requirements, and the present disclosure embodiment does not limit the specific value of the set polling period.

[0124] In the embodiments of the present disclosure, the user can also actively query the execution status of various tasks through the user terminal. For example, after the user submits a new task on the user terminal, the user can click to query the task execution result on the task processing interface corresponding to the new task, and the user terminal submits the task request information for querying the execution status query task in response to the user's query operation. Or, when the user switches from the task processing interface of the current task to the task processing interface of another task, a task for querying the execution status of the task corresponding to the switched task processing interface can be automatically generated. Or, when the user switches from the current task processing interface to the task processing interface of another task, the user can also click to query the task execution result on the switched task processing interface, and the user terminal submits the task request information for querying the execution status query task to the server in response to the user's query operation.

[0125] S520. Generate a model interaction request in real time according to the task request information.

[0126] Correspondingly, step S520 may specifically include the following operations:

[0127] S521. Obtain the task query identifier used by the execution status query task to query the task to be queried according to the task request information.

[0128] Among them, the task to be queried may be the task whose execution status needs to be queried by the execution status query task.

[0129] After the server obtains the task request information of the execution status query task, it can obtain the task query identifier included in the task request information to query the execution status of the task to be queried that needs to be queried by the execution status query task according to the task query identifier.

[0130] S522. Determine the second target model for processing the task to be queried according to the task query identifier.

[0131] Among them, the second target model may be the model for processing the task to be queried.

[0132] S523. Generate the task status acquisition request in real time according to the second target model and the task query identifier.

[0133] Among them, the task status acquisition request may be a request for requesting the model server to feedback the task execution status of the task to be queried.

[0134] Specifically, the server can query the stored identification mapping relationship according to the task query identification to determine the task identification matching the to-be-query task, so as to determine the to-be-query task whose status needs to be queried by the client according to the queried task identification, and determine the second target model that is currently processing the to-be-query task according to the task query identification, thereby generating a task status acquisition request matching the execution status query task in real time as a real-time model interaction request.

[0135] According to the above technical solution, by generating a task status acquisition request in real time for the task request information of the execution status query task submitted by the client, the execution status query task of the user can be responded to in real time, and the processing efficiency of the execution status query task can be improved.

[0136] S530. Send the task status acquisition request to the model server in real time.

[0137] S540. Receive the current execution status of the to-be-processed task feedback by the model server in real time according to the model task status acquisition request.

[0138] Wherein, the current execution status of the to-be-processed task may be the current execution status of the to-be-query task.

[0139] Correspondingly, after the server generates the task status acquisition request, it can send the task status acquisition request to the model server in real time. After receiving the task status acquisition request, the model server can determine the to-be-query task to be queried by the execution status query task and the second target model for executing the to-be-query task according to the task status acquisition request, so as to obtain the current model processing result of the second target model for the to-be-query task as the current execution status of the to-be-processed task, and send the obtained current execution status of the to-be-processed task to the server.

[0140] S550. Determine the current task processing result of the to-be-processed task according to the current model processing result.

[0141] Correspondingly, step S550 may specifically include the following operations:

[0142] S551. Judge whether the current execution status of the to-be-processed task is the task execution in progress status. If so, execute S552; otherwise, execute S553.

[0143] S552. Generate the task running in processing result of the to-be-query task queried by the execution status query task.

[0144] Wherein, the task running in processing result may be that the to-be-query task is still in execution, that is, the result that has not been processed yet.

[0145] S553. Determine that the execution status of the current task to be processed is the task execution completed status, and receive the task execution result of the to-be-query task queried by the execution status query task feedback from the model server.

[0146] Among them, the task execution result can be the final execution result of the to-be-query task.

[0147] After the server receives the execution status of the current task to be processed, it can judge the execution status of the current task to be processed. If it is determined that the execution status of the current task to be processed is the task execution in progress status, indicating that the to-be-query task has not been processed yet, the task running in progress processing result of the to-be-query task can be generated as the current task processing result. If it is determined that the execution status of the current task to be processed is the task execution completed status, indicating that the to-be-query task has been processed, the task execution result of the to-be-query task feedback from the model server can be received as the current task processing result, such as receiving the task result data of the to-be-query task, and the task result data can be, for example, the image recognition result or the calculated target project plan, etc.

[0148] Correspondingly, after the model server feedbacks the execution status of the current task to be processed, or after feedbacking the task execution result, the connection with the server can be disconnected.

[0149] S560. Feedback the task running in progress processing result or the task execution result to the user terminal.

[0150] Correspondingly, after the server generates the task running in progress processing result or the task execution result, the task running in progress processing result or the task execution result can be feedback to the user terminal, and the user terminal can then obtain the query result of the execution status query task in real time.

[0151] Adopting the above technical solution, after the user terminal submits the execution status query task, it disconnects from the server. The server generates a task status acquisition request for interacting with the second target model according to the execution status query task, and requests the model server to feedback the current execution status of the second target model independently processing the to-be-query task without connection in real time through the task status acquisition request, and feedbacks the queried status to the user terminal, which can avoid the problem of task processing failure caused by network jitter delay, and the problem of task processing failure caused by service crash due to large thread overhead and connection overhead, so as to ensure the success rate of task execution and improve the real-time performance and efficiency of task processing.

[0152] In an example, Figure 6It is a flowchart of a task processing method provided by an embodiment of the present disclosure. Based on the technical solutions of the above embodiments, the embodiment of the present disclosure has been optimized and improved. In the case where the task to be processed is a deletion task, multiple specific and optional implementation manners are given for obtaining task request information sent by the client according to the task to be processed, generating a model interaction request in real time according to the task request information, receiving the current model processing result feedback by the model server in real time according to the model interaction request, and determining the current task processing result of the task to be processed according to the current model processing result.

[0153] As Figure 6 shown, a task processing method includes:

[0154] S610. Obtain task request information sent by the client according to the deletion task.

[0155] Among them, the deletion task may be a task for deleting a task that the model is currently processing and executing.

[0156] In the embodiment of the present disclosure, in addition to submitting a new task and querying the status of the currently executing task, the client can also perform a deletion operation on the task that the model is currently processing and executing to instruct the model to cancel the processing of the currently executing task, thereby submitting a deletion task. When the user initiates a deletion task matching a certain task to be deleted through the client, the server can receive the task request information sent by the client according to the deletion task. After the client submits the task request information of the deletion task, the connection with the server can be disconnected.

[0157] S620. Generate a model interaction request in real time according to the task request information.

[0158] Correspondingly, step S620 may specifically include the following operations:

[0159] S621. Obtain a task query identifier used by the deletion task to delete the task to be deleted according to the task request information.

[0160] Among them, the task to be deleted may be the task that the deletion task requests to delete, and this task may be the task that the model is currently processing.

[0161] When the server obtains the task request information of the deletion task, it can obtain the task query identifier included in the task request information to query the task to be deleted that the deletion task needs to delete according to the task query identifier.

[0162] S622. Determine a third target model for processing the task to be deleted according to the task query identifier.

[0163] Among them, the third target model may be the model that is currently processing the task to be deleted.

[0164] S623. Generate the task deletion request in real time according to the third target model and the task query identifier.

[0165] Among them, the task deletion request can be a request for requesting the model server to delete the task to be deleted.

[0166] Specifically, the server can query the stored identifier mapping relationship according to the task query identifier to determine the task identifier matching the task to be queried, so as to determine the task to be deleted that the client needs to delete according to the queried task identifier, and determine the third target model that is currently processing the task to be deleted according to the task query identifier, so as to generate a task deletion request matching the deletion task in real time as a real-time model interaction request.

[0167] In the above technical solution, by generating a task deletion request in real time for the task request information of the deletion task submitted by the client, the deletion task of the user can be responded to in real time, and the processing efficiency of the deletion task can be improved.

[0168] S630. Send the task deletion request to the model server in real time.

[0169] S640. Receive the task deletion result of the task to be deleted that the model server feeds back in real time according to the task deletion request.

[0170] Among them, the task deletion result can be the deletion result of the task to be deleted.

[0171] S650. Generate the deletion task response data for deleting the task to be deleted according to the task deletion result of the task to be deleted.

[0172] Among them, the deletion task response data can be the response data generated by the server according to the task deletion result.

[0173] S660. Feed back the deletion task response data for deleting the task to be deleted to the client.

[0174] In the embodiment of the present disclosure, after the server receives the task deletion result of the task to be deleted that the model server feeds back in real time, it can further generate the deletion task response data for deleting the task to be deleted according to the task deletion result of the task to be deleted. Exemplarily, if the task deletion result is that the task has been processed and cannot be deleted, the server can generate the deletion task response data indicating deletion failure. If the task deletion result is that the task has been deleted, the server can generate the deletion task response data indicating deletion success. Correspondingly, the server can feed back the generated deletion task response data to the client in real time so that the client can obtain the deletion result of the deletion task.

[0175] With the above technical solution, after the client submits a deletion task, it disconnects from the server. The server generates a task deletion request for interacting with the third target model according to the deletion task, and uses the task deletion request to request the model server to feedback the result of deleting the task to be deleted in real time, and generates a deletion task response data to feedback to the client. This can avoid the problem of task processing failure caused by network jitter delay, and the problem of task processing failure caused by service crash due to large thread overhead and connection overhead, thus ensuring the success rate of task execution and improving the real-time performance and efficiency of task processing.

[0176] In one example, Figure 7 is a flowchart of a task processing method provided by an embodiment of the present disclosure. The embodiment of the present disclosure gives a schematic diagram of the interaction process between the client, the server, and the model server for processing various tasks.

[0177] Correspondingly, as Figure 7 shown, the user can query the server through the client whether there is a task in execution that the user is operating on currently. If it exists, the server can return specific task information; if not, it can return an empty result.

[0178] If the client queries that there is no task in execution that the user is operating on currently, a new task can be created, for example, by clicking the new query on the client to create a new task, and submitting the task configuration information of the new task. The server can call the model service to process the new task according to the new task and the task configuration information. After receiving the new task, the model server can schedule the model to process the new task and return the task number (i.e., the task identifier) of the new task to the server. At this time, the server can establish a mapping relationship between the task query identifier sent by the client and the task identifier, and locally store the task query identifier, the task identifier, and the identifier mapping relationship between the two. At this time, since the model server has not completed processing the new task, the server does not need to feedback the result to the client.

[0179] Correspondingly, if the user stays on the task processing interface of the newly created task to wait for the execution result of the task after submitting the new task, there is no need for the user side to maintain a long connection with the server side, and between the server side and the model server side. However, to avoid the user waiting for a long time, the user side can periodically poll the server to query the execution result of the task in the current task processing interface. After receiving the task request information for querying the execution status sent by the user side, the server side can generate a task status acquisition request in real time and send it to the model server side. The model server side can obtain the execution result of the task in the current task processing interface in real time according to the task status acquisition request and feedback it to the server side. The server side can then generate the current task processing result in real time according to the result feedback by the model server side and feedback it to the user side, such as prompting the user that the task is being executed or feeding back the execution result data of the task to the user.

[0180] After the user submits a new task, the user can switch to the task processing interface to perform the processing operations of other tasks. For example, the user can switch to another task processing interface to query the execution status of the corresponding task. Similarly, if the user does not perform other operations after switching the task processing interface, the user side can periodically poll the server to query the execution result of the task in the current task processing interface.

[0181] The user can also cancel the currently executing task through the user side. Specifically, when the user submits a delete task for the currently executing task through the user side, the server side can generate a task deletion request matching the delete task and send it to the model server side. The model server side can then respond to the task deletion request, promptly delete the task that the user needs to delete, and feedback the result data of the delete task to the user side in real time. The server side can then feedback the execution result of the delete task to the user side according to the result data of the delete task.

[0182] In an optional embodiment of the present disclosure, the task processing method may further include: obtaining the task storage association data of the to-be-processed task; storing the task storage association data in a database; wherein, the task storage association data includes at least one of the following: the task configuration information of the newly created task, the task identifier feedback by the model server side, and the task status of the to-be-processed task.

[0183] Among them, the task storage association data may be the relevant data of the to-be-processed task that the server side needs to store.

[0184] Such as Figure 7As shown in the figure, when the server interacts with the user terminal and the model server, the server can store the task storage association data of the task to be processed in the database according to the interaction result. Optionally, the server can store it in the database in real time after obtaining the task configuration information of the task to be processed sent by the user terminal. At the same time, after the server receives the task identifier of the task to be processed fed back by the model server, it can also be stored in the database. When the model server starts to process the task to be processed, it can record the task information processed by the model server in the database. At the same time, after the model server finishes executing the task to be processed, it can store the task execution result of the task to be processed in the database in real time. At the same time, if the user terminal initiates a task deletion, when the task status to be deleted changes, the server can also update the task status in the database.

[0185] Through the above technical solution, by storing the task storage association data in the database in real time, the storage backup record of the data can be realized, which is beneficial to tracing the root cause of abnormal problems.

[0186] In an optional embodiment of the present disclosure, the model server can be used to respond to the model interaction request in real time by adopting the method of distributed scheduling of model operator workers and the method of distributed computing.

[0187] In order to realize the de-service of the model server, the model operator can be independently used as a work to execute. A model can be completed by multiple model operator workers together. Therefore, the model server can adopt the method of distributed scheduling of model operator workers to complete the model scheduling function, so as to connect to a more suitable computing platform according to the actual business scenario and improve the scalability and adaptability of task processing. At the same time, the model server can also adopt the method of distributed computing to respond to the model interaction request sent by the server in real time to make full use of the cluster computing power of the model server.

[0188] By adopting the above technical solution, the model operator is scheduled to process the task to be processed through an asynchronous model scheduling method. The user does not need to stay on the waiting page for a long time. During the model calculation, the user can perform other operations or return to the page at any time to view the task execution result. At the same time, the user terminal can maintain a historical task list to help the user review the running results of the historical tasks configured before, without repeated calls to execute, reducing computing power loss. The asynchronous system interaction mode can also reduce the coupling between service modules, the module functions are clearer, and abandoning the long connection real-time response method can improve service stability. Since the execution method of the model operator is more diversified, the distributed scheduling of the operator can be realized, the de-service of the worker, and the atomicity of the single worker function. In the big data computing scenario, the computing efficiency can be improved by means of a distributed computing platform.

[0189] In one exampleFigure 8 It is a structural diagram of a task processing device provided by an embodiment of the present disclosure. The embodiments of the present disclosure are applicable to the situation of using a real-time interaction method to call a model to process a to-be-processed task of a client interaction. The device is implemented by software and / or hardware and is specifically configured in an electronic device. The electronic device can be a server device and is used in cooperation with a client and a model server.

[0190] As Figure 8 shown, a task processing device 800 includes: a task request information acquisition module 810, a model interaction request generation module 820, a model interaction request sending module 830, a model current processing result receiving module 840, and a current task processing result determination module 850. Among them,

[0191] The task request information acquisition module 810 is configured to acquire task request information sent by the client according to the to-be-processed task;

[0192] The model interaction request generation module 820 is configured to generate a model interaction request in real time according to the task request information;

[0193] The model interaction request sending module 830 is configured to send the model interaction request to the model server in real time;

[0194] The model current processing result receiving module 840 is configured to receive the model current processing result feedback by the model server in real time according to the model interaction request;

[0195] The current task processing result determination module 850 is configured to determine the current task processing result of the to-be-processed task according to the model current processing result.

[0196] After the embodiments of the present disclosure acquire the task request information sent by the client according to the to-be-processed task, generate a model interaction request in real time according to the acquired task request information, and send the real-time generated model interaction request to the model server in real time, so that the model server processes the to-be-processed task according to the model interaction request, and receives the model current processing result feedback by the model server in real time according to the model interaction request, thereby determining the current task processing result of the to-be-processed task according to the received model current processing result, solving the problems of low real-time performance and efficiency of task processing in the related art when processing tasks according to the model, and being able to improve the real-time performance and efficiency of task processing.

[0197] Optionally, the task to be processed includes a new task; the model interaction request includes a model scheduling request; the model interaction request generation module 820 is further configured to: obtain the task configuration information of the new task according to the task request information; determine a first target model for processing the new task according to the task configuration information; and generate the model scheduling request in real time according to the first target model.

[0198] Optionally, the model current processing result receiving module 840 is further configured to: receive the task identifier that is fed back in real time by the model server according to the model scheduling request; the current task processing result determining module 850 is further configured to: obtain the task query identifier sent by the user side according to the new task; establish and store the identification mapping relationship between the task query identifier and the task identifier; and generate the current task processing result according to the task query identifier, the task identifier, and the identification mapping relationship.

[0199] Optionally, the current task processing result determining module 850 is further configured to: obtain the space factor identifier, the user identifier, and the model identifier sent by the user side according to the task to be processed; and generate the task query identifier according to the space factor identifier, the user identifier, and the model identifier.

[0200] Optionally, the task to be processed includes an execution status query task; the model interaction request includes a model task status acquisition request; the task request information acquisition module 810 is further configured to: obtain the task request information of the execution status query task sent by the user side at a set polling period; and / or obtain the task request information of the execution status query task sent by the user side in response to a user query operation; the model interaction request generation module 820 is further configured to: obtain the task query identifier for querying the task to be queried for the execution status query task according to the task request information; determine a second target model for processing the task to be queried according to the task query identifier; and generate the task status acquisition request in real time according to the second target model and the task query identifier.

[0201] Optionally, the model current processing result receiving module 840 is further configured to: receive the current to-be-processed task execution status that is fed back in real time by the model server according to the model task status acquisition request; the current task processing result determination module 850 is further configured to: when determining that the current to-be-processed task execution status is the in-task execution status, generate the task running processing result of the to-be-query task queried by the execution status query task; or, when determining that the current to-be-processed task execution status is the task execution completed status, receive the task execution result of the to-be-query task queried by the execution status query task fed back by the model server; the task processing device further includes: a task execution result feedback module, configured to feed back the task running processing result or the task execution result to the user terminal.

[0202] Optionally, the to-be-processed task includes a deletion task; the model interaction request includes a task deletion request; the model interaction request generation module 820 is further configured to: obtain, according to the task request information, a task query identifier of the deletion task for deleting the to-be-deleted task; determine a third target model for processing the to-be-deleted task according to the task query identifier; and generate the task deletion request in real time according to the third target model and the task query identifier.

[0203] Optionally, the model current processing result receiving module 840 is further configured to: receive the task deletion result of the to-be-deleted task that is fed back in real time by the model server according to the task deletion request; the current task processing result determination module 850 is further configured to: generate deletion task response data for deleting the to-be-deleted task by the deletion task according to the task deletion result of the to-be-deleted task; the task processing device further includes: a deletion task response data feedback module, configured to feed back the deletion task response data for deleting the to-be-deleted task by the deletion task to the user terminal.

[0204] Optionally, the task processing device further includes: a task storage associated data acquisition module, configured to acquire task storage associated data of the to-be-processed task; a task storage associated data storage module, configured to store the task storage associated data into a database; where the task storage associated data includes at least one of the following: task configuration information of a new task, task information processed by the model server, a task identifier fed back by the model server, and the task status of the to-be-processed task.

[0205] Optionally, the model server is configured to respond to the model interaction request in real time by adopting a method of distributing and scheduling model operator workers and a method of distributed computing.

[0206] The above task processing device can execute the task processing method provided in any embodiment of the present disclosure, and has the corresponding functional modules and beneficial effects of the execution method. For technical details not described in detail in this embodiment, reference may be made to the task processing method provided in any embodiment of the present disclosure.

[0207] Since the above-described task processing device is a device that can execute the task processing method in the embodiments of the present disclosure, based on the task processing method described in the embodiments of the present disclosure, those skilled in the art can understand the specific implementation manners and various variations of the task processing device in this embodiment. Therefore, the details of how the task processing device implements the task processing method in the embodiments of the present disclosure will not be described in detail here. As long as the device adopted by those skilled in the art to implement the task processing method in the embodiments of the present disclosure belongs to the scope protected by the present disclosure.

[0208] In one example, Figure 9 is a structural diagram of a task processing system provided by an embodiment of the present disclosure. The embodiments of the present disclosure are applicable to the situation of using a real-time interaction method to call a model to process a to-be-processed task of a user-side interaction, such as Figure 9 As shown, the task processing system 900 may include a user side 910, a server side 920, and a model server side 930; the user side 910 is communicatively connected to the server side 920, and the server side 920 is communicatively connected to the model server side 930; where:

[0209] The user side 910 is configured to generate task request information according to the to-be-processed task and send the task request information to the server side 920;

[0210] The server side 920 is configured to generate a model interaction request in real time according to the task request information and send the model interaction request to the model server side 930 in real time;

[0211] The model server side 910 is configured to feedback the current processing result of the model in real time according to the model interaction request and send the current processing result of the model to the server side 920;

[0212] The server side 920 is further configured to determine the current task processing result of the to-be-processed task according to the current processing result of the model.

[0213] After obtaining the task request information sent by the client according to the task to be processed in an embodiment of the present disclosure, a model interaction request is generated in real time according to the obtained task request information, and the real-time generated model interaction request is sent to the model server in real time, so that the model server processes the task to be processed according to the model interaction request, and receives the current model processing result feedback by the model server in real time according to the model interaction request, thereby determining the current task processing result of the task to be processed according to the received current model processing result, solving the problems of low real-time performance and efficiency of task processing in related technologies when processing tasks according to the model, and being able to improve the real-time performance and efficiency of task processing.

[0214] Optionally, the task to be processed includes a new task; the model interaction request includes a model scheduling request; the server 920 is further configured to: obtain the task configuration information of the new task according to the task request information; determine a first target model for processing the new task according to the task configuration information; and generate the model scheduling request in real time according to the first target model.

[0215] Optionally, the model server 930 is further configured to: feedback a task identifier to the server 920 in real time according to the model scheduling request; the server 920 is further configured to: obtain a task query identifier sent by the client 910 according to the new task; establish and store an identifier mapping relationship between the task query identifier and the task identifier; and generate the current task processing result according to the task query identifier, the task identifier, and the identifier mapping relationship.

[0216] Optionally, the server 920 is further configured to: obtain a space factor identifier, a user identifier, and a model identifier sent by the client according to the task to be processed; and generate the task query identifier according to the space factor identifier, the user identifier, and the model identifier.

[0217] Optionally, the task to be processed includes an execution status query task; the model interaction request includes a model task status acquisition request; the client 910 is further configured to: send the task request information of the execution status query task to the server 920 according to a set polling period; and / or, send the task request information of the execution status query task to the server 920 in response to a user query operation; obtain a task query identifier used by the execution status query task to query the task to be queried according to the task request information; determine a second target model for processing the task to be queried according to the task query identifier; and generate the task status acquisition request in real time according to the second target model and the task query identifier.

[0218] Optionally, the model server 930 is further configured to: in real time, according to the request for the model task status, feed back the execution status of the current task to be processed to the server 920; the server 920 is further configured to: in the case where it is determined that the execution status of the current task to be processed is the in-task execution status, generate the processing result during the task execution of the task to be queried for the execution status query task; or, in the case where it is determined that the execution status of the current task to be processed is the task execution completed status, receive the task execution result of the task to be queried for the execution status query task fed back by the model server; and feed back the processing result during the task execution or the task execution result to the user terminal.

[0219] Optionally, the task to be processed includes a deletion task; the model interaction request includes a task deletion request; the server 920 is further configured to: according to the task request information, obtain the task query identifier of the deletion task for deleting the task to be deleted; determine the third target model for processing the task to be deleted according to the task query identifier; and in real time, generate the task deletion request according to the third target model and the task query identifier.

[0220] Optionally, the model server 930 is further configured to: in real time, according to the task deletion request, feed back the task deletion result of the task to be deleted to the server 920; the server 920 is further configured to: generate the deletion task response data for deleting the task to be deleted according to the task deletion result of the task to be deleted; and feed back the deletion task response data for deleting the task to be deleted to the user terminal.

[0221] Optionally, the server 920 is further configured to: obtain the task storage association data of the task to be processed; and store the task storage association data in the database; wherein, the task storage association data includes at least one of the following: the task configuration information of the new task, the task identifier fed back by the model server, and the task status of the task to be processed.

[0222] Optionally, the model server 930 is configured to respond to the model interaction request in real time by adopting the method of distributed scheduling of model operator workers and the method of distributed computing.

[0223] Figure 10 is a schematic diagram of the architecture of a task processing system provided by an embodiment of the present disclosure. In a specific example, as Figure 10 shown, the user terminal can support the user to configure task configuration information in a visual interface, such as asset allocation information such as a portfolio list, a benchmark type, a currency type, a penetration type, and a date. After the task configuration information is configured, when the user clicks "Run", it enters as Figure 11The task processing interface shown is in the loading state, waiting for the model side to complete the calculation. At the same time, the user side can add a polling logic to query the model calculation result. The user side can poll once every 5s. For the user's perceived maximum latency time is 5s, so as to ensure that the user can be timely informed after some models with faster calculations are completed. During the calculation, the user side does not need to maintain a connection with the server. A new task can be uniquely identified by the spaceId, userId, and moduleId. When the user returns to this task processing interface after operating on other task processing interfaces, the previously configured task status can still be queried.

[0224] The server can maintain historical model tasks through a database. By persisting the task status and results, the traceability of historical tasks is ensured. In addition to providing a model task interface to the user side, the server can also provide an additional function for querying the task status, and judge whether the user side is in a waiting state or directly render the model calculation result through the task status. At the same time, the server can also perform data format conversion and cleaning on certain specific data structures to adapt to the input data structure of the model. The server can also maintain user operation records, regularly record user behaviors for problem tracing. The server can also monitor the interface performance to ensure the quality of data interaction with the user side and the model server. The server can also connect to the model scheduling layer interface and adapt the scheduler to schedule the model worker to execute according to the business scenario.

[0225] The model scheduling layer of the model server is an interface service provided by the model layer to the outside, and provides the following interfaces:

[0226] (1) Submit a model calculation task: responsible for receiving task configuration parameters, verification, and scheduling resources to execute the model calculation logic;

[0227] (2) Check the task status: obtain the calculation status of the corresponding model through a unique task identifier, and for models that have been executed successfully or failed, the upper layer can be timely informed of the calculation results;

[0228] (3) Forcefully end the task: The currently executing model task can be forcefully ended through this interface to release the server pressure, and the user can reconfigure and execute the model calculation.

[0229] Figure 12 is a schematic diagram of an architecture for the server to adapt to the model server for distributed scheduling provided by an embodiment of the present disclosure, Figure 13 is a schematic diagram of a process for the Airflow scheduling model worker to execute tasks provided by an embodiment of the present disclosure, Figure 14 is a schematic diagram of a process for Spark distributed computing provided by an embodiment of the present disclosure. In a specific example, such as Figure 12 ,Figure 13 and Figure 14 As shown in Figure 14 , the model scheduling layer can introduce Airflow (task scheduler) and Gearman (queue-based task scheduler, distributed task scheduler) to achieve distributed task scheduling, ensuring the reasonable utilization of cluster computing resources. For multiple relatively complex and strongly dependent models, the execution of the pipeline for scheduling the models can be configured through workflow (task scheduler). Different workers can be scheduled to execute different pending tasks, and multiple workers can also be scheduled to execute the same pending task. When the computing pressure is too high, the built-in task queue of Gearman can achieve task waiting and diversion. At the same time, Airflow can support the Spark (task scheduler) operator (controller), and Spark jobs can be directly submitted and run through BashOperator or SparkSubmitOperator, supporting distributed computing, improving the computing rate, reducing the user waiting time, and providing a convenient and fast solution for the big data computing scenario of the model.

[0230] For the big data model computing scenario in the model server's model computing layer, Hadoop (distributed system infrastructure) and Spark can be introduced to achieve distributed model computing, making full use of the cluster computing power and greatly improving the computing efficiency. Models with relatively small computing amounts can still reuse the previous logic and be provided externally through the unified interface layer, and different computing solutions can be implemented according to different business scenarios.

[0231] The model operator layer of the model server extracts the computing logic from the previous interface-based service and forms independent workers according to the model type. The functions of the workers are atomized, and a single model is designed as a worker and scheduled by the scheduling layer for execution. After the functions of the workers are lightweight, the computing logic can be simplified, facilitating scheduling and execution, making full use of computing resources. At the same time, for models with inclusive and similar functions, the computing logic can be reused, the overall structure is clearer, and the scalability and pluggability are stronger.

[0232] It can be seen that in the above task processing system, the interaction mode between the user side and the model side is changed from synchronous to asynchronous. After submitting the model calculation task, the user does not need to stay on the page for a long time to wait and can execute other task types synchronously, thereby improving the task processing efficiency and user experience. The model side is deserviced, and the model operators are independently executed as workers, and the calculation solutions are diversified, and more suitable calculation platforms can be docked according to the actual business scenarios. Distributed task scheduling systems such as Airflow and Gearman are introduced to effectively regulate the model calculation pressure under high-pressure conditions. For the calculation scenarios of big data, Hadoop and Spark are introduced to implement distributed computing jobs, which greatly improves the calculation efficiency at the software level. The above task processing system is an optimization at the completely software level, which optimizes the model calculation method and the model interaction method, greatly improves the system stability, and also greatly improves the calculation efficiency, and can reduce the hardware implementation cost.

[0233] In the technical solution of the present disclosure, the processing of the collection, storage, use, processing, transmission, provision, and disclosure of the user's personal information involved all complies with the provisions of relevant laws and regulations and does not violate public order and good customs.

[0234] It should be noted that any permutation and combination of the technical features in the above embodiments also belong to the protection scope of the present disclosure.

[0235] In one example, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0236] Figure 15 FIG. shows a schematic block diagram of an example electronic device 1500 that can be used to implement the embodiments of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0237] As Figure 15As shown, device 1500 includes a computing unit 1501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1502 or a computer program loaded from a storage unit 1508 into a random access memory (RAM) 1503. In the RAM 1503, various programs and data required for the operation of the device 1500 can also be stored. The computing unit 1501, the ROM 1502, and the RAM 1503 are connected to each other via a bus 1504. An input / output (I / O) interface 1505 is also connected to the bus 1504.

[0238] Multiple components in the device 1500 are connected to the I / O interface 1505, including: an input unit 1506, such as a keyboard, a mouse, etc.; an output unit 1507, such as various types of displays, speakers, etc.; a storage unit 1508, such as a magnetic disk, an optical disk, etc.; and a communication unit 1509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1509 allows the device 1500 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0239] The computing unit 1501 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 1501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 1501 executes the various methods and processes described above, such as the task processing method. For example, in some embodiments, the task processing method can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 1508. In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 1500 via the ROM 1502 and / or the communication unit 1509. When the computer program is loaded into the RAM 1503 and executed by the computing unit 1501, one or more steps of the task processing method described above can be executed. Alternatively, in other embodiments, the computing unit 1501 can be configured to execute the task processing method in any other appropriate manner (e.g., by means of firmware).

[0240] The various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that receives data and instructions from a storage system, at least one input device, and at least one output device, and transmits the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0241] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program codes are executed by the processor or controller, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, executed partially on the machine as an independent software package and partially on a remote machine, or executed entirely on a remote machine or server.

[0242] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include electrical connections based on one or more wires, 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), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0243] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, speech input, or tactile input).

[0244] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0245] A computer system can include a client and a server. The client and the server are generally far from each other and usually interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services. The server can also be a server of a distributed system, or a server combined with a blockchain.

[0246] After obtaining the task request information sent by the user side according to the task to be processed in an embodiment of the present disclosure, a model interaction request is generated in real time according to the obtained task request information, and the real-time generated model interaction request is sent to the model server in real time, so that the model server processes the task to be processed according to the model interaction request, and receives the current processing result of the model fed back in real time by the model server according to the model interaction request, thereby determining the current task processing result of the task to be processed according to the received current processing result of the model, solving the problems such as low real-time performance and efficiency of task processing existing in the related art when processing tasks according to the model, and being able to improve the real-time performance and efficiency of task processing.

[0247] It should be understood that the various forms of processes shown above can be used, with steps reordered, added or deleted. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and no limitations are imposed herein.

[0248] The above specific embodiments do not constitute a limitation on the protection scope of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub - combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the protection scope of this disclosure.

Claims

1. A task processing method, comprising: Obtaining task request information sent by a client according to a task to be processed; wherein, the task to be processed includes a new task; Generating a model interaction request in real time according to the task request information; wherein, the model interaction request includes a model scheduling request; Sending the model interaction request to a model server in real time; Receiving a current processing result of the model feedback by the model server in real time according to the model interaction request; Determining a current task processing result of the task to be processed according to the current processing result of the model; Wherein, the receiving the current processing result of the model feedback by the model server in real time according to the model interaction request includes: Receiving a task identifier feedback by the model server in real time according to the model scheduling request; The determining the current task processing result of the task to be processed according to the current processing result of the model includes: Obtaining a task query identifier sent by the client according to the new task; Establishing and storing an identifier mapping relationship between the task query identifier and the task identifier; Generating the current task processing result according to the task query identifier, the task identifier, and the identifier mapping relationship; The obtaining the task query identifier sent by the client according to the task to be processed includes: Obtaining a space factor identifier, a user identifier, and a model identifier sent by the client according to the task to be processed; Generating the task query identifier according to the space factor identifier, the user identifier, and the model identifier.

2. The method according to claim 1, wherein, The generating a model interaction request in real time according to the task request information includes: Obtaining task configuration information of the new task according to the task request information; Determining a first target model for processing the new task according to the task configuration information; Generating the model scheduling request in real time according to the first target model.

3. The method according to claim 1, wherein, The task to be processed further includes an execution status query task; the model interaction request further includes a model task status obtaining request; The obtaining task request information sent by the client according to the task to be processed includes: Obtaining the task request information of the execution status query task sent by the client at a set polling period; and / or Obtaining the task request information of the execution status query task sent by the client in response to a user query operation; The generating a model interaction request in real time according to the task request information includes: Obtaining a task query identifier used by the execution status query task to query a task to be queried according to the task request information; Determining a second target model for processing the task to be queried according to the task query identifier; Generating the task status obtaining request in real time according to the second target model and the task query identifier.

4. The method according to claim 3, wherein The receiving the current processing result of the model feedback by the model server in real time according to the model interaction request includes: Receiving the current execution status of the task to be processed feedback by the model server in real time according to the model task status obtaining request; The determining the current task processing result of the task to be processed according to the current processing result of the model includes: When it is determined that the execution status of the current task to be processed is the in - progress task execution status, generate the processing result during task execution of the task to be queried in the execution status query task query; or When it is determined that the execution status of the current task to be processed is the task execution completed status, receive the task execution result of the task to be queried in the execution status query task query fed back by the model server; The method further includes: Feed back the processing result during task execution or the task execution result to the client.

5. The method according to claim 1, wherein The task to be processed further includes a deletion task; the model interaction request further includes a task deletion request; The generating the model interaction request in real - time according to the task request information includes: Obtain the task query identifier of the deletion task for deleting the task to be deleted according to the task request information; Determine the third target model for processing the task to be deleted according to the task query identifier; Generate the task deletion request in real - time according to the third target model and the task query identifier.

6. The method according to claim 5, wherein, The receiving the current model processing result fed back by the model server in real - time according to the model interaction request includes: Receive the task deletion result of the task to be deleted fed back by the model server in real - time according to the task deletion request; The determining the current task processing result of the task to be processed according to the current model processing result includes: Generate the deletion task response data for deleting the task to be deleted by the deletion task according to the task deletion result of the task to be deleted; The method further includes: Feed back the deletion task response data for deleting the task to be deleted by the deletion task to the client.

7. The method according to claim 1, further includes: Obtain the task storage associated data of the task to be processed; Store the task storage associated data in the database; Wherein, the task storage associated data includes at least one of the following: the task configuration information of the new task, the task identifier fed back by the model server, and the task status of the task to be processed.

8. The method according to claim 1, wherein The model server is used to respond to the model interaction request in real - time by adopting the method of distributed scheduling of model operator worker and the method of distributed computing.

9. A task processing device, including: A task request information acquisition module, configured to acquire the task request information sent by the client according to the task to be processed; wherein, the task to be processed includes a new task; A model interaction request generation module, configured to generate a model interaction request in real - time according to the task request information; wherein, the model interaction request includes a model scheduling request; A model interaction request sending module, configured to send the model interaction request to the model server in real - time; A current model processing result receiving module, configured to receive the current model processing result fed back by the model server in real - time according to the model interaction request; A current task processing result determination module, configured to determine the current task processing result of the task to be processed according to the current model processing result; Wherein, the current model processing result receiving module is further used for: Receive the task identifier fed back by the model server in real - time according to the model scheduling request; The current task processing result determination module is further configured to: Obtain a task query identifier sent by the client according to the newly created task; Establish and store an identifier mapping relationship between the task query identifier and the task identifier; Generate the current task processing result according to the task query identifier, the task identifier, and the identifier mapping relationship; The current task processing result determination module is further configured to: Obtain a space factor identifier, a user identifier, and a model identifier sent by the client according to the task to be processed; Generate the task query identifier according to the space factor identifier, the user identifier, and the model identifier.

10. The apparatus according to claim 9, wherein, The model interaction request generation module is further configured to: Obtain the task configuration information of the newly created task according to the task request information; Determine a first target model for processing the newly created task according to the task configuration information; Generate the model scheduling request in real time according to the first target model.

11. The apparatus according to claim 9, wherein, The task to be processed further includes an execution status query task; the model interaction request further includes a model task status acquisition request; the task request information acquisition module is further configured to: Obtain the task request information of the execution status query task sent by the client at a set polling period; and / or Obtain the task request information of the execution status query task sent by the client in response to a user query operation; The model interaction request generation module is further configured to: Obtain a task query identifier used by the execution status query task to query the task to be queried according to the task request information; Determine a second target model for processing the task to be queried according to the task query identifier; Generate the task status acquisition request in real time according to the second target model and the task query identifier.

12. The apparatus according to claim 11, wherein, The model current processing result receiving module is further configured to: Receive the current execution status of the task to be processed in real time fed back by the model server according to the model task status acquisition request; The current task processing result determination module is further configured to: Generate a task running processing result of the task to be queried queried by the execution status query task when it is determined that the current execution status of the task to be processed is the task execution in progress state; or Receive the task execution result of the task to be queried queried by the execution status query task fed back by the model server when it is determined that the current execution status of the task to be processed is the task execution completed state; The apparatus further includes: a task execution result feedback module, configured to feedback the task running processing result or the task execution result to the client.

13. A task processing system, including a client, a server, and a model server; the client is communicatively connected to the server, and the server is communicatively connected to the model server; wherein: The client is configured to generate task request information according to a task to be processed, and send the task request information to the server; wherein, the task to be processed includes a newly created task; The server is used to generate a model interaction request in real time according to the task request information, and send the model interaction request to the model server in real time; wherein, the model interaction request includes a model scheduling request. The model server is used to feedback the current processing result of the model in real time according to the model interaction request, and feedback the current processing result of the model to the server. The server is further used to determine the current task processing result of the to-be-processed task according to the current processing result of the model. Wherein, the model server is further used to: feedback the task identifier to the server in real time according to the model scheduling request; the server is further used to: obtain the task query identifier sent by the user terminal according to the new task; establish and store the identification mapping relationship between the task query identifier and the task identifier; generate the current task processing result according to the task query identifier, the task identifier and the identification mapping relationship. The server is further used to: obtain the space factor identifier, user identifier and model identifier sent by the user terminal according to the to-be-processed task; generate the task query identifier according to the space factor identifier, the user identifier and the model identifier.

14. An electronic device, comprising: At least one processor; And A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the task processing method according to any one of claims 1-8.

15. A non-transitory computer-readable storage medium storing computer instructions, the computer instructions being used to cause a computer to execute the task processing method according to any one of claims 1-8.

16. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by a processor, the task processing method according to any one of claims 1-8 is implemented.

Images