Space-time task processing method and system and storage medium

Abstract

The embodiment of the invention discloses a space-time task processing method and system and a storage medium, and is applied to the technical field of artificial intelligence information processing. The method comprises the steps of splitting a to-be-processed space-time task into a text reasoning sub-task and a space-time calculation sub-task, generating a programmed thinking chain based on the text reasoning sub-task and the space-time calculation sub-task, determining a calculation tool combination required by the space-time calculation sub-task, and further calculating the space-time task according to the calculation tool combination and the programmed thinking chain. And generating a standardized execution tool for calling the calculation tool combination in the process of executing the programmed thinking chain, namely executing the standardized execution tool in the process of executing the programmed thinking chain to obtain a processing result of the space-time task to be processed. In this way, the defects that an existing space-time task processing system cannot have space-time calculation such as dynamic modeling deficiency, cross-domain fusion deficiency and numerical reasoning errors in space-time modal data cognition are overcome, and dynamic and cross-domain complex space-time tasks can be processed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence information processing, and in particular to a spatiotemporal task processing method and system and a storage medium. BACKGROUND

[0002] Large language models (LLMs) are a type of deep learning algorithm that has the ability to recognize, condense, translate, predict, and generate text and other types of content. LLMs are an important category of transformer models that have proven to have very broad uses, as they not only enable artificial intelligence systems to use human language, but also can be applied in various fields, such as healthcare, software code writing, and many other multi-faceted tasks.

[0003] Specifically, when processing spatiotemporal modal tasks, LLM models can mainly use traditional Chain of Thought (CoT) methods that can linearly decompose tasks, Program-Assisted Language (PAL) methods, and traditional spatiotemporal data analysis methods such as GIS-based spatiotemporal reasoning methods, but it is difficult to deal with some complex spatiotemporal tasks with dynamic and cross-domain characteristics. SUMMARY

[0004] The embodiments of the present application provide a spatiotemporal task processing method, system and storage medium, which realize processing of complex spatiotemporal tasks with dynamic and cross-domain characteristics.

[0005] In one aspect, the embodiments of the present application provide a spatiotemporal task processing method, comprising: obtaining a to-be-processed spatiotemporal task, and splitting the to-be-processed spatiotemporal task into a text reasoning subtask and a spatiotemporal calculation subtask; generating a programmatic chain of thought based on the text reasoning subtask and the spatiotemporal calculation subtask; determining a combination of calculation tools required for executing the spatiotemporal calculation subtask in the programmatic chain of thought; generating a standardized execution tool that calls the combination of calculation tools in the process of executing the programmatic chain of thought according to the combination of calculation tools and the programmatic chain of thought; executing the standardized execution tool in the process of executing the programmatic chain of thought to obtain a processing result of the to-be-processed spatiotemporal task.

[0006] In another aspect, the embodiments of the present application provide a spatiotemporal task processing system, comprising: a task splitting unit configured to obtain a to-be-processed spatiotemporal task, and split the to-be-processed spatiotemporal task into a text reasoning subtask and a spatiotemporal calculation subtask; a thinking chain unit configured to generate a programmed thinking chain based on the text reasoning subtask and the space-time calculation subtask; a tool determination unit configured to determine a calculation tool combination required for performing the space-time calculation subtask in the programmed thinking chain; an execution tool generation unit configured to generate, according to the calculation tool combination and the programmed thinking chain, a standardized execution tool for invoking the calculation tool combination in the process of executing the programmed thinking chain; a processing unit configured to execute the standardized execution tool in the process of executing the programmed thinking chain, to obtain a processing result of the to-be-processed space-time task.

[0007] Another aspect of the embodiment of the present application also provides a computer readable storage medium storing a plurality of computer programs, the computer programs being adapted to be loaded by a processor and execute the space-time task processing method according to the aspect of the embodiment of the present application.

[0008] It can be seen that, in the method of the embodiment, the to-be-processed space-time task is split into a text reasoning subtask and a space-time calculation subtask, a programmed thinking chain based on the text reasoning subtask and the space-time calculation subtask is generated, a calculation tool combination required for the space-time calculation subtask is determined, and then, according to the calculation tool combination and the programmed thinking chain, a standardized execution tool for invoking the calculation tool combination in the process of executing the programmed thinking chain is generated, so that the standardized execution tool can be executed in the process of executing the programmed thinking chain, to obtain a processing result of the to-be-processed space-time task. In this way, the two types of subtasks involved in the to-be-processed space-time task, i.e., the space-time calculation subtask and the text reasoning subtask, are decoupled, the space-time calculation subtask is implemented by using another calculation tool combination, the defects of the existing space-time task processing system in space-time modal data cognition, such as dynamic modeling deficiency, cross-domain fusion deficiency and numerical reasoning error, are made up, and the complex space-time task with dynamic and cross-domain characteristics can be processed. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0010] Figure 1 is a schematic diagram of a space-time task processing method provided by an embodiment of the present application; Figure 2 is a flowchart of a space-time task processing method provided by an embodiment of the present application; Figure 3 is a schematic diagram of a spatiotemporal task processing system in an application embodiment of the present application; Figure 4 is a flowchart of a spatiotemporal task processing method provided in an application embodiment of the present application; Figure 5 is a logical structure schematic diagram of a spatiotemporal task processing system provided in an embodiment of the present application; Figure 6 is a logical structure schematic diagram of a terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0012] The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein, for example, can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0013] The embodiments of the present application provide a spatiotemporal task processing method, mainly a method for processing a relatively complex spatiotemporal task, as shown in Figure 1 The spatiotemporal task processing system mainly implements the processing of a to-be-processed spatiotemporal task according to the following steps: obtaining a to-be-processed spatiotemporal task, splitting the to-be-processed spatiotemporal task into a text reasoning subtask and a spatiotemporal calculation subtask, generating a programmed thinking chain based on the text reasoning subtask and the spatiotemporal calculation subtask, determining a calculation tool combination required for executing the spatiotemporal calculation subtask in the programmed thinking chain, generating a standardized execution tool for calling the calculation tool combination in the process of executing the programmed thinking chain according to the calculation tool combination and the programmed thinking chain, and executing the standardized execution tool in the process of executing the programmed thinking chain to obtain a processing result of the to-be-processed spatiotemporal task.

[0014] The spatiotemporal task processing system in this embodiment can be an LLM model or the like.

[0015] One embodiment of the present application provides a spatiotemporal task processing method, mainly a method executed by a spatiotemporal task processing system, a flowchart of which is shown in Figure 2 as follows. Step 101: obtaining a to-be-processed spatiotemporal task, and splitting the to-be-processed spatiotemporal task into a text reasoning subtask and a spatiotemporal calculation subtask.

[0016] It can be understood that a user can input a to-be-processed spatiotemporal task through interaction with the spatiotemporal task processing system, and when the spatiotemporal task processing system obtains the to-be-processed spatiotemporal task, it needs to first perform intent recognition on the to-be-processed spatiotemporal task to identify which specific subtasks need to be executed to achieve the to-be-processed spatiotemporal task. In this embodiment, there are mainly two types of subtasks, namely a text reasoning subtask and a spatiotemporal calculation subtask. The text reasoning subtask is mainly a task implemented through a reasoning process of text information, and the spatiotemporal calculation subtask is mainly a task implemented through a certain spatial calculation.

[0017] For example, for a to-be-processed spatiotemporal task “simulation of rainstorm water depth in a certain area”, the text reasoning subtask can be “rainstorm return period reasoning”, and the spatiotemporal calculation subtask can be “hydrological model iterative calculation”.

[0018] In the specific implementation process, when the to-be-processed spatiotemporal task is obtained: The spatiotemporal analysis memory such as Retrieval-Augmented Generation (RAG) can be combined to retrieve spatiotemporal data associated with the to-be-processed spatiotemporal task; An intent recognition module is used to perform intent recognition on the to-be-processed spatiotemporal task to extract core constraint conditions of the to-be-processed spatiotemporal task; A programmed thinking strong prompt word template such as a Plan Template module is called to split the to-be-processed spatiotemporal task into a text reasoning subtask and a spatiotemporal calculation subtask, and to determine spatiotemporal data and core constraint conditions corresponding to each subtask in the text reasoning subtask and the spatiotemporal calculation subtask from the retrieved spatiotemporal data and the extracted core constraint conditions.

[0019] The programmed thinking strong prompt word template mainly includes task type labeling, spatiotemporal calculation stripping rules, and tool calling formats. The task type labeling can be used to label the types of each subtask included in the to-be-processed spatiotemporal task, and thus the text reasoning subtask and the spatiotemporal calculation subtask can be split.

[0020] Step 102, generating a programmed thinking chain based on the text reasoning subtask and the space-time calculation subtask.

[0021] The process of generating a programmed thinking chain is to decompose a complex space-time task to be processed into a series of clear, executable and linear steps. In the embodiment, the generated programmed thinking chain can include a plurality of executable steps arranged in a certain order, and the plurality of executable steps mainly include execution steps based on the text reasoning subtask and the space-time calculation subtask.

[0022] Specifically, the programmed thinking chain can be generated by the above-mentioned called programmed thinking strong prompt word template.

[0023] Step 103, determining a calculation tool combination required for executing the space-time calculation subtask in the programmed thinking chain.

[0024] Since mathematical calculation is required when executing the space-time calculation subtask, and the space-time task processing system itself generally does not have the function of executing mathematical calculation, it is necessary to first search for a calculation tool combination used for executing the space-time calculation subtask in an external system, such as a spatial positioning tool, a time interpolation tool, a multi-modal fusion tool, etc. If the corresponding calculation tool combination is not found from the external system, the calculation tool required for executing the space-time calculation subtask can also be directly created. Specifically: Discovering a plurality of external candidate tools required for the space-time calculation subtask through a model context protocol (MCP) gateway; Selecting an optimal tool combination for executing the space-time calculation subtask from the plurality of external candidate tools as the calculation tool combination, and obtaining the calling parameter format of each calculation tool in the optimal tool combination; If the model context protocol gateway does not discover the external candidate tool required for executing the space-time calculation subtask, creating a model context protocol-based calculation tool required for executing the space-time calculation subtask.

[0025] Among them, the execution of the space-time calculation subtask can be realized by executing a plurality of unit tasks, and each unit task can have a plurality of selectable tools, i.e., a plurality of external candidate tools. Therefore, a selectable tool needs to be selected for each unit task, and the selectable tools corresponding to the plurality of unit tasks form the calculation tool combination for executing the space-time calculation subtask. In a specific implementation process, the selected selectable tools need to form a calculation tool combination that is optimal in terms of calculation overhead. Specifically: Determining a plurality of tool sets required for executing the space-time calculation subtask from the plurality of external candidate tools; Calculating the calculation overhead of each tool set in executing the space-time calculation subtask; The tool set with the minimum calculation overhead is selected as the calculation tool combination.

[0026] The spatial positioning tool, the time interpolation tool, and the multi-modal fusion tool in the plurality of external candidate tools are sorted in ascending order of calculation overhead as follows: the spatial positioning tool, the time interpolation tool, and the multi-modal fusion tool.

[0027] At step 104, a standardized execution tool that invokes the calculation tool combination during execution of the programmed thinking chain is generated according to the calculation tool combination and the programmed thinking chain.

[0028] Specifically, for the execution step of invoking the calculation tool combination in the programmed thinking chain, a corresponding execution tool that conforms to the model context protocol standard, i.e., a standardized execution tool, is generated. In this way, during execution of the programmed thinking chain, an external tool executor can be invoked to execute the execution tool that conforms to the model context protocol standard, thereby achieving invocation of the calculation tool combination.

[0029] At step 105, the standardized execution tool is executed during execution of the programmed thinking chain to obtain a processing result of the to-be-processed spatio-temporal task.

[0030] It can be understood that, during execution of the programmed thinking chain, when the standardized execution tool is executed by the external tool executor, the following can be performed: The execution process of the standardized execution tool is monitored; If error critical information, such as spatio-temporal coordinate calculation errors, is monitored during execution of the standardized execution tool, the error critical information is returned to execute step 101 to re-split the to-be-processed spatio-temporal task into a text reasoning subtask and a spatio-temporal calculation subtask, correct the text reasoning subtask and the spatio-temporal calculation subtask obtained based on the to-be-processed spatio-temporal task intention recognition, and re-execute steps 101 to 105 on the corrected text reasoning subtask and spatio-temporal calculation subtask; In this way, steps 101 to 105 are repeatedly executed to split the to-be-processed spatio-temporal task, generate a programmed thinking chain, determine a calculation tool combination, generate a standardized execution tool, and execute the standardized execution tool, to obtain a processing result of the to-be-processed spatio-temporal task after multiple cycles.

[0031] Through the above-mentioned repeated execution of steps 101 to 105, a feedback loop mechanism is formed, which effectively repairs the standardized execution tool generated during processing of the to-be-processed spatio-temporal task, and improves the accuracy of processing of the to-be-processed spatio-temporal task, in particular, the accuracy of processing of the spatio-temporal calculation subtask contained in the to-be-processed control task.

[0032] As can be seen, in the method of the embodiment, the to-be-processed spatio-temporal task is split into a text reasoning subtask and a spatio-temporal computing subtask, and a programmed thinking chain based on the text reasoning subtask and the spatio-temporal computing subtask is generated, the computing tool combination required by the spatio-temporal computing subtask is determined, and then the standardized execution tool that calls the computing tool combination in the process of executing the programmed thinking chain is generated according to the computing tool combination and the programmed thinking chain. In this way, the two types of subtasks involved in the to-be-processed spatio-temporal task, i.e., the spatio-temporal computing subtask and the text reasoning subtask, are decoupled, and another computing tool combination is used to implement the spatio-temporal computing subtask, which makes up for the defects of the existing spatio-temporal task processing system in dynamic modeling, cross-domain fusion, and numerical reasoning error in spatio-temporal modal data cognition, and can implement processing of complex spatio-temporal tasks with dynamic and cross-domain characteristics.

[0033] The following will illustrate the spatio-temporal task processing method of the application with a specific application example. The method in the embodiment can be applied to a spatio-temporal task processing system as shown in Figure 3 The spatio-temporal task processing system of the embodiment can be an LLM model, which mainly includes the following parts: task splitting 11, directional acyclic graph (DAG) automated tool calling 12, standardized execution tool generation 13, and result evaluation and fitting 14, wherein: The task splitting 11 is mainly used to split the to-be-processed spatio-temporal task input by the user into a text reasoning subtask and a spatio-temporal computing subtask, and generate a programmed thinking chain.

[0034] The DAG automated tool calling 12 is mainly used to find the computing tool of the MCP gateway based on the ordered scheduling of the DAG automated construction of the computing tool.

[0035] The standardized execution tool generation 13 is mainly used to generate a standardized execution tool conforming to the MCP protocol under the joint cooperation of the PlanTemplate module and the programmed thinking chain, and start the standardized execution tool conforming to the MCP protocol in the external tool interpreter.

[0036] The result evaluation and fitting 14 is mainly used to monitor the execution process of the standardized execution tool, find the key error information in the execution process, and provide the key error information to the user.

[0037] Specifically, as shown in Figure 4 The spatio-temporal task processing system of the embodiment can implement processing of a more complex to-be-processed spatio-temporal task according to the following steps: Step 201, the user inputs a to-be-processed spatio-temporal task, such as "storm water depth simulation in a certain area", through interaction with the spatio-temporal task processing system.

[0038] Specifically, the task splitting 11 in the spatio-temporal task processing system in this embodiment can retrieve spatio-temporal data associated with the to-be-processed spatio-temporal task in combination with the RAG, such as retrieving terrain elevation and pipe network layout; and retrieve key parameters associated with the to-be-processed spatio-temporal task in combination with the historical waterlogging event library, such as rainfall intensity of 100 mm / h and pipe network drainage capacity of 500 cubic meters / s.

[0039] Further, intent recognition is performed on the to-be-processed spatio-temporal task, and it is determined that the intent of the to-be-processed spatio-temporal task is to capture waterlogging features through multi-modal data fusion, and then extract core constraint conditions corresponding to a spatio-temporal calculation subtask in the to-be-processed spatio-temporal task, such as a spatial range of 1.2 square kilometers and a time window of 3 hours.

[0040] Step 202, the task splitting 11 splits the to-be-processed spatio-temporal task into a text reasoning subtask and a spatio-temporal calculation subtask, and generates a programmatic thinking chain based on the text reasoning subtask and the spatio-temporal calculation subtask.

[0041] Specifically, the PlanTemplate module can be called to split the to-be-processed spatio-temporal task "storm water depth simulation in a certain area" into a text reasoning subtask "storm return period reasoning" and a spatio-temporal calculation subtask "hydrological model iterative calculation", and generate a programmatic thinking chain containing the text reasoning subtask and the spatio-temporal calculation subtask based on the PlanTemplate module, wherein the programmatic thinking chain can be generated in combination with a program template library when generating the programmatic thinking chain.

[0042] Step 203, the MCP gateway is used to externally search for external candidate tools required for executing the spatio-temporal calculation subtask, such as hydrological analysis tools and Digital Elevation Model (DEM) spatial interpolation tools, to construct a DAG workflow; if no corresponding calculation tool is found from the outside, a calculation tool corresponding to the above-mentioned spatio-temporal calculation subtask that conforms to the MCP protocol needs to be created in real time, and tool invocation parameters are obtained.

[0043] The external candidate tools that can be searched for from the outside through the MCP gateway can include spatio-temporal semantic analysis tools, hydrological analysis service tools, spatio-temporal deduction tools, 3D modeling tools, and the like.

[0044] Step 204, when the MCP gateway finds the external candidate tool, the optimal tool combination for executing the spatio-temporal calculation subtask can be obtained from the external candidate tool through the DAG automation tool call 12, as the calculation tool combination for executing the above-mentioned spatio-temporal calculation subtask, and the order of the ordered scheduling of the calculation tools in the optimal tool combination is determined.

[0045] Step 205, the standardized execution tool generation 13 can generate a standardized execution tool conforming to the MCP protocol according to the above-mentioned calculation tool combination, which is mainly used to call the calculation tool combination in the execution of the above-mentioned programmed thinking chain process, and then call the MCP gateway through the external tool interpreter to complete the self-start of the standardized execution tool, so as to dynamically link the city twin world model service to obtain real-time data of the meteorological bureau.

[0046] Step 206, when the standardized execution tool is started, in the execution process of the above-mentioned programmed thinking chain, when the spatio-temporal calculation subtask is executed, the above-mentioned standardized execution tool can be executed, which is mainly realized by calling the related calculation tools through the MCP gateway, and then the processing result corresponding to the to-be-processed spatio-temporal task is obtained.

[0047] It should be noted that in the process of executing the programmed thinking chain, the intermediate results used in the calculation process can be transmitted through gRPC streaming, and the processing result of the above-mentioned “rainstorm water depth simulation in a certain area” to-be-processed spatio-temporal task can be visualized as a water depth field (resolution 0.5m x 0.5m) on the digital twin base, and a structured risk prompt visualization (maximum water depth 0.8m / duration 45 minutes) can also be output.

[0048] Step 207, in the process of executing the standardized execution tool, the execution process of the standardized execution tool is monitored through result evaluation and fitting 14, key error information in the execution process is found, and the key error information is fed back to the user, and the above-mentioned steps 202 to 207 are returned based on the key error information.

[0049] In this way, the above-mentioned steps 202 to 207 can be executed in a loop, so that the processing result of the to-be-processed spatio-temporal task obtained is more accurate.

[0050] For example, in the process of repeatedly executing the above-mentioned steps 201 to 207 on the above-mentioned to-be-processed spatio-temporal task “rainstorm water depth simulation in a certain area”, the Manning coefficient (n value is adjusted from 0.013 to 0.018) is re-calibrated by linking the historical waterlogging records of the water bureau, and after multiple iterations, the simulation error is reduced from 23% to 7%, and a more accurate waterlogging tracing report is finally generated.

[0051] It can be seen that the spatio-temporal task processing method of the embodiment can realize: By leveraging the inherent synergistic mechanism between spatial modality cognition and procedural thinking, this approach compensates for the shortcomings of Large Language Models (LLM) in the practical application of spatiotemporal modality data cognition.

[0052] By decoupling the text reasoning subtask and the spatiotemporal computation subtask in the spatiotemporal task to be processed, the PoT (Programming Thinking) technology is integrated with the theory of spatial intelligence, providing a new technical path for the construction of interpretable spatial intelligent systems.

[0053] The MCP gateway can help intelligent agents discover spatiotemporal modality cognition tools on their own, providing a set of common components for handling complex spatiotemporal tasks.

[0054] This invention also provides a spatiotemporal task processing system, the structural schematic of which is shown below. Figure 5 As shown, it can specifically include: The task splitting unit 20 is used to acquire the spatiotemporal task to be processed and split the spatiotemporal task to be processed into a text reasoning subtask and a spatiotemporal calculation subtask.

[0055] The task splitting unit 20 is specifically used to retrieve spatiotemporal data associated with the spatiotemporal task to be processed through spatiotemporal analysis memory; perform intent recognition on the spatiotemporal task to be processed to extract the core constraints of the spatiotemporal task to be processed; call the procedural thinking strong prompt word template to split the spatiotemporal task to be processed into a text reasoning subtask and a spatiotemporal calculation subtask, and determine the spatiotemporal data and core constraints corresponding to the text reasoning subtask and the spatiotemporal calculation subtask respectively from the spatiotemporal data and core constraints.

[0056] The thought chain unit 21 is used to generate a programmatic thought chain based on the text reasoning subtask and spatiotemporal calculation subtask split by the task splitting unit 20.

[0057] Tool determination unit 22 is used to determine the combination of computing tools required to execute the spatiotemporal computing subtask in the procedural thinking chain generated by the thinking chain unit 21.

[0058] The tool determination unit 22 is specifically used to discover multiple external candidate tools required for the spatiotemporal computation subtask through the model context protocol gateway; select the optimal tool combination for executing the spatiotemporal computation subtask from the multiple external candidate tools as the computation tool combination; if the model context protocol gateway does not discover external candidate tools required for executing the spatiotemporal computation subtask, create a computation tool based on the model context protocol required for executing the spatiotemporal computation subtask.

[0059] The plurality of external candidate tools include a plurality of spatial positioning tools, time interpolation tools, and multi-modal fusion tools. When the tool determination unit 22 selects an optimal tool combination for performing the spatio-temporal calculation subtask from the plurality of external candidate tools as the calculation tool combination, the tool determination unit 22 is specifically configured to determine a plurality of tool sets required for performing the spatio-temporal calculation subtask from the plurality of external candidate tools; calculate the calculation overhead of each tool set when performing the spatio-temporal calculation subtask, respectively; and select a tool set with the minimum calculation overhead as the calculation tool combination.

[0060] The execution tool generation unit 23 is configured to generate a standardized execution tool that calls the calculation tool combination in the process of executing the programmed thinking chain according to the calculation tool combination and the programmed thinking chain determined by the tool determination unit 22.

[0061] The execution tool generation unit 23 is specifically configured to generate an execution tool that conforms to the model context protocol standard for an execution step that calls the calculation tool combination in the programmed thinking chain.

[0062] The processing unit 24 is configured to execute the standardized execution tool generated by the execution tool generation unit 23 in the process of executing the programmed thinking chain to obtain a processing result of the to-be-processed spatio-temporal task.

[0063] Further, the spatio-temporal task processing system of the embodiment can further include: The monitoring unit 25 is configured to monitor the execution process of the standardized execution tool, and if error critical information in the process of executing the standardized execution tool is monitored, the task splitting unit 20 is informed to re-execute the step of splitting the to-be-processed spatio-temporal task into a text reasoning subtask and a spatio-temporal calculation subtask according to the error critical information.

[0064] In this way, the system of the embodiment can repeatedly execute the steps of splitting the to-be-processed spatio-temporal task, generating the programmed thinking chain, determining the calculation tool combination, generating the standardized execution tool, and executing the standardized execution tool through the monitoring of the execution process of the standardized execution tool by the monitoring unit 25, to obtain a processing result of the to-be-processed spatio-temporal task after multiple cycles.

[0065] In the system of the embodiment, the two types of subtasks involved in the to-be-processed spatio-temporal task, i.e., the spatio-temporal calculation subtask and the text reasoning subtask, are decoupled, and another calculation tool combination is used to implement the spatio-temporal calculation subtask, which makes up for the defects of the existing spatio-temporal task processing system, such as the lack of dynamic modeling, insufficient cross-domain fusion, and numerical reasoning errors in spatio-temporal modal data cognition, and can implement the processing of complex spatio-temporal tasks with dynamics and cross-domain.

[0066] An embodiment of the present application further provides a terminal device, a structural schematic diagram of which is shown in Figure 6 The terminal device can have great differences due to different configurations or performances, and can include one or more central processing units (CPUs) 30 (for example, one or more processors) and a memory 31, one or more storage media 32 (for example, one or more mass storage devices) storing application programs 321 or data 322. The memory 31 and the storage media 32 can be temporary storage or persistent storage. The programs stored in the storage media 32 can include one or more modules (not shown in the figure), each of which can include a series of instruction operations in the terminal device. Further, the central processing unit 30 can be configured to communicate with the storage media 32 and execute the series of instruction operations in the storage media 32 on the terminal device.

[0067] Specifically, the application programs 321 stored in the storage media 32 include application programs for spatio-temporal task processing, and the programs can include the task splitting unit 20, the thought chain unit 21, the tool determining unit 22, the execution tool generating unit 23, the processing unit 24 and the monitoring unit 25 in the spatio-temporal task processing system, which will not be described herein. Further, the central processing unit 30 can be configured to communicate with the storage media 32 and execute a series of operations corresponding to the application programs for spatio-temporal task processing stored in the storage media 32 on the terminal device.

[0068] The terminal device can further include one or more power supplies 33, one or more wired or wireless network interfaces 34, one or more input / output interfaces 35, and / or one or more operating systems 323, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

[0069] The steps performed by the spatio-temporal task processing system in the method embodiments described above can be based on the Figure 6 The structure of the terminal device is shown in the figure.

[0070] Further, an embodiment of the present application further provides a computer readable storage medium storing a plurality of computer programs, the computer programs being adapted to be loaded and executed by a processor to perform the spatio-temporal task processing method performed by the spatio-temporal task processing system.

[0071] An embodiment of the present application further provides a terminal device including a processor and a memory; The memory is configured to store a plurality of computer programs for loading and execution by the processor to implement the spatio-temporal task processing method performed by the spatio-temporal task processing system; and the processor is configured to implement each of the plurality of computer programs.

[0072] Those skilled in the art can understand that all or part of the steps in the above-mentioned various methods of the embodiments can be completed by instructing the relevant hardware through a program, and the program can be stored in a computer readable storage medium, which can include read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0073] The above has carried on the detailed introduction to the spatio-temporal task processing method, system and storage medium provided by the embodiments of the present application, the principle and implementation mode of the present application are described in this paper by applying specific examples; the above embodiment is only used to help understand the method and core idea of the present application; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed; in view of the above, the content of the specification should not be understood as the limitation of the present application.

Claims

1. A method for spatiotemporal task processing, characterized by, The method comprises: acquiring a to-be-processed spatio-temporal task, and splitting the to-be-processed spatio-temporal task into a text reasoning subtask and a spatio-temporal calculation subtask; generating a programmed thinking chain based on the text reasoning subtask and the spatio-temporal calculation subtask; determining a calculation tool combination required for executing the spatio-temporal calculation subtask in the programmed thinking chain; generating a standardized execution tool for calling the calculation tool combination in the process of executing the programmed thinking chain according to the calculation tool combination and the programmed thinking chain; executing the standardized execution tool in the process of executing the programmed thinking chain to obtain a processing result of the to-be-processed spatio-temporal task.

2. The method of claim 1, wherein, The splitting of the to-be-processed spatio-temporal task into the text reasoning subtask and the spatio-temporal calculation subtask specifically comprises: retrieving spatio-temporal data associated with the to-be-processed spatio-temporal task through a spatio-temporal analysis memory; performing intent recognition on the to-be-processed spatio-temporal task to extract core constraint conditions of the to-be-processed spatio-temporal task; calling a programmed thinking strong prompt word template to split the to-be-processed spatio-temporal task into the text reasoning subtask and the spatio-temporal calculation subtask, and determining spatio-temporal data and core constraint conditions corresponding to the text reasoning subtask and the spatio-temporal calculation subtask from the spatio-temporal data and the core constraint conditions.

3. The method of claim 1, wherein, The determination of the calculation tool combination required for executing the spatio-temporal calculation subtask in the programmed thinking chain specifically comprises: discovering a plurality of external candidate tools required for the spatio-temporal calculation subtask through a model context protocol gateway; selecting an optimal tool combination for executing the spatio-temporal calculation subtask from the plurality of external candidate tools as the calculation tool combination.

4. The method of claim 3, wherein, The plurality of external candidate tools comprise a plurality of spatial positioning tools, time interpolation tools, and multi-modal fusion tools. The selection of the optimal tool combination for executing the spatio-temporal calculation subtask from the plurality of external candidate tools as the calculation tool combination specifically comprises: determining a plurality of tool sets required for executing the spatio-temporal calculation subtask from the plurality of external candidate tools; calculating a calculation overhead of each tool set in executing the spatio-temporal calculation subtask; selecting a tool set with the smallest calculation overhead as the calculation tool combination.

5. The method of claim 3, wherein, The determination of the calculation tool combination required for executing the spatio-temporal calculation subtask in the programmed thinking chain further comprises: if the model context protocol gateway does not discover an external candidate tool required for executing the spatio-temporal calculation subtask, creating a model context protocol-based calculation tool required for executing the spatio-temporal calculation subtask.

6. The method according to any one of claims 1 to 5, characterized in that, The generation of the standardized execution tool for calling the calculation tool combination in the process of executing the programmed thinking chain according to the calculation tool combination and the programmed thinking chain specifically comprises: for an execution step of calling the calculation tool combination in the programmed thinking chain, generating a corresponding execution tool conforming to a model context protocol standard.

7. The method according to any one of claims 1 to 5, wherein The method further comprises: monitoring an execution process of the standardized execution tool; If error critical information is monitored in the execution of the standardized execution tool process, the step of splitting the to-be-processed spatio-temporal task into a text reasoning subtask and a spatio-temporal calculation subtask is returned to be re-executed according to the error critical information.

8. The method of claim 7, wherein, The splitting of the to-be-processed spatio-temporal task, the generation of the programmatic thinking chain, the determination of the calculation tool combination, the generation of the standardized execution tool, and the execution of the standardized execution tool are performed multiple times to obtain a processing result of the to-be-processed spatio-temporal task after multiple cycles.

9. A spatio-temporal task processing system, characterized by Comprise: a task splitting unit configured to acquire a to-be-processed spatio-temporal task and split the to-be-processed spatio-temporal task into a text reasoning subtask and a spatio-temporal calculation subtask; a thinking chain unit configured to generate a programmatic thinking chain based on the text reasoning subtask and the spatio-temporal calculation subtask; a tool determination unit configured to determine a calculation tool combination required for executing a spatio-temporal calculation subtask in the programmatic thinking chain; an execution tool generation unit configured to generate, according to the calculation tool combination and the programmatic thinking chain, a standardized execution tool that invokes the calculation tool combination in the execution of the programmatic thinking chain; a processing unit configured to execute the standardized execution tool in the execution of the programmatic thinking chain to obtain a processing result of the to-be-processed spatio-temporal task.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a plurality of computer programs adapted to be loaded and executed by the processor to perform the spatio-temporal task processing method according to any one of claims 1 to 8.

Images