Panorama chain monitoring method and reproduction method of artificial intelligence application platform

Abstract

The application discloses a panoramic chain monitoring method and reproduction method of an artificial intelligence application platform; the method comprises the following steps: acquiring business processing information, calling each application module according to the business processing information, each application module being a module which has been automatically instrumented according to a link development kit and business configuration information; in response to detecting that an instrumented link in the application module starts an application program interface (API) to be called, generating a link identifier, generating node information of a link node in the link according to the calling of the link node recorded by the instrumented link, and the node information comprising a node identifier, a parent node identifier corresponding to the link node and the link identifier; in response to detecting that an instrumented link in the application module ends the API to be called, stopping the recording of the calling of the link node in the link, and generating link information according to the node information of the recorded link node, so that the problems of low link monitoring efficiency and easy errors are solved; and the automatic instrumenting is realized, so that the efficiency is higher and the accuracy is higher.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a panoramic chain monitoring method and a reproduction method for an artificial intelligence application platform. Background Technology

[0002] AI application platforms integrate different modules, forming AI application products with long chains, complex logic, and spanning multiple applications. Existing technical solutions typically focus on a single business or app, with relatively simple logical structures and an emphasis on the independence and specificity of modules. This leads to the need for manual integration of upstream and downstream interfaces of business modules and manual instrumentation to link business chains when dealing with constantly changing requirements. However, the IDs of upstream and downstream links may be inconsistent; for example, some use request_id and track_id, while others use trace_id. Especially when dealing with concurrent operations from multiple applications and multiple threads, manual association is prone to errors, consuming significant resources and time, resulting in low efficiency and becoming a weakness. It cannot adapt to the needs of rapidly iterating AI platforms and makes it difficult to achieve complete and efficient monitoring of the entire chain. Summary of the Invention

[0003] This invention provides a panoramic chain monitoring method and a reproduction method for an artificial intelligence application platform to solve problems such as low efficiency and error susceptibility in chain monitoring.

[0004] According to one aspect of the present invention, a panoramic chain monitoring method for an artificial intelligence application platform is provided, the artificial intelligence application platform including at least one application module, the method comprising:

[0005] Obtain business processing information, and call each of the application modules according to the business processing information, wherein each of the application modules is a module that has been automatically embedded according to the link development toolkit and business configuration information;

[0006] In response to the detection that the link start application programming interface (API) of the embedded point in the application module is called, a link identifier is generated. According to the call of the link node in the link recorded by the embedded point, the node information of the link node is generated according to the link identifier. The node information includes the node identifier, the parent node identifier corresponding to the link node, and the link identifier.

[0007] In response to the detection that the link end API of the embedded point in the application module is called, the recording of the link node calls in the link is stopped, and link information is generated based on the node information of the recorded link node.

[0008] According to another aspect of the present invention, a method for reproducing the entire chain of an artificial intelligence application platform is provided, the artificial intelligence application platform including at least one application module, the method comprising:

[0009] Obtain link information, which is generated based on the node information of the link nodes. The node information includes a node identifier, the parent node identifier corresponding to the link node, and a link identifier. The link nodes are recorded based on the data points of each application module. Each application module automatically performs data point recording based on the link development toolkit and business configuration information.

[0010] The link is reproduced based on the link information.

[0011] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0012] At least one processor, and a memory communicatively connected to said at least one processor;

[0013] The memory stores a computer program that can be executed by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to execute the panoramic chain monitoring method or the panoramic chain reproduction method of the artificial intelligence application platform according to any embodiment of the present invention.

[0014] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a processor to execute and implement the panoramic chain monitoring or reproduction method of the artificial intelligence application platform described in any embodiment of the present invention.

[0015] According to another aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program, which, when executed by a processor, implements the panoramic chain monitoring or reproduction method of the artificial intelligence application platform described in any embodiment of the present invention.

[0016] This invention provides a panoramic chain monitoring method for an artificial intelligence application platform. The artificial intelligence application platform includes at least one application module. The method acquires business processing information and calls each application module based on the business processing information. Each application module is a module that has been automatically embedded with tracking points based on a link development toolkit and business configuration information. In response to detecting a link start application programming interface (API) call in the application module, a link identifier is generated. Based on the tracking points, the method records calls to link nodes in the link and generates node information for each link node based on the link identifier. The node information includes a node identifier, a parent node identifier corresponding to the link node, and a link identifier. In response to detecting a link end API call in the application module, the method stops recording calls to link nodes in the link and generates a link based on the recorded node information. This information system addresses the issues of low efficiency and error-proneness in link monitoring. By using a link development toolkit and business configuration information to embed tracking points in various application modules, it achieves automated tracking, eliminating the need for manual operation, resulting in higher efficiency, greater accuracy, and reduced errors, significantly improving development efficiency. It acquires business processing information and calls application modules based on this information. Link nodes are automatically collected based on pre-set tracking points. During the link node collection process, when the API for starting a link in an application module is called, it triggers the generation of a panoramic link. First, a link identifier is generated. Then, based on the tracking records of link node calls, node information is generated based on the link identifier. This node information includes the node identifier, parent node identifier, and link identifier. The node information allows for accurate link reconstruction, enabling effective panoramic link monitoring of the AI ​​application platform.

[0017] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a panoramic chain monitoring method for an artificial intelligence application platform provided according to Embodiment 1 of the present invention;

[0020] Figure 2This is a flowchart of a panoramic chain monitoring method for an artificial intelligence application platform according to Embodiment 2 of the present invention;

[0021] Figure 3 This is an example diagram of a vehicle infotainment system configuration service provided according to Embodiment 2 of the present invention;

[0022] Figure 4 This is an example diagram illustrating the connection between an AI application platform and the entire supply chain according to Embodiment 2 of the present invention;

[0023] Figure 5 This is an example diagram of a link and data flow provided according to Embodiment 2 of the present invention;

[0024] Figure 6 This is an example diagram of a link reporting method provided according to Embodiment 2 of the present invention;

[0025] Figure 7 This is an example diagram illustrating a non-real-time data reporting implementation according to Embodiment 2 of the present invention;

[0026] Figure 8 This is a flowchart of a panoramic chain reproduction method for an artificial intelligence application platform according to Embodiment 3 of the present invention;

[0027] Figure 9 This is a system architecture diagram for panoramic chain execution monitoring and reproduction provided in Embodiment 3 of the present invention;

[0028] Figure 10 This is a schematic diagram illustrating an implementation example of data acquisition and processing according to Embodiment 3 of the present invention;

[0029] Figure 11 This is a schematic diagram of the structure of the panoramic chain monitoring device of the artificial intelligence application platform provided in Embodiment 4 of the present invention;

[0030] Figure 12 This is a schematic diagram of the structure of a panoramic chain reproduction device for an artificial intelligence application platform according to Embodiment 5 of the present invention;

[0031] Figure 13 This is a schematic diagram of the structure of an electronic device provided in Embodiment Six of the present invention. Detailed Implementation

[0032] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0034] Example 1

[0035] Figure 1 This is a flowchart of a panoramic chain monitoring method for an artificial intelligence application platform provided in Embodiment 1 of the present invention. This embodiment is applicable to the situation of panoramic chain monitoring of an artificial intelligence application platform. The method can be executed by a panoramic chain monitoring device of the artificial intelligence application platform. The panoramic chain monitoring device of the artificial intelligence application platform can be implemented in hardware and / or software form, and can be configured in a vehicle; the artificial intelligence application platform includes at least one application module. Figure 1 As shown, the method includes:

[0036] S101. Obtain business processing information and call each application module according to the business processing information. Each application module is a module that has been automatically marked with data points according to the link development toolkit and business configuration information.

[0037] In this embodiment, the application module can be an AI main interaction, an information aggregation framework, a generative user interface (UI), a task processing unit (TPU), or functional components (e.g., atomic services or atomic capabilities). The TPU completes the atomic capability management framework, promotes business atomicity, connects the entire link, and automatically provides edge-side atomic capabilities to algorithms. Business processing information can be understood as business data that needs to be processed through an AI application platform, such as user voice, gestures, and operation information collected during human-computer interaction, or multimodal input data, contextual information from large language models, model generation results, etc. The link development toolkit can be understood as a toolkit for developing different types of links. The link development toolkit can be a Trace SDK, responsible for defining a unified data protocol for the entire link, implementing the core of upstream and downstream link connectivity, collecting data, and providing a flexible and unified API for business users to call. Service configuration information can be understood as information for configuring services. Service configuration information can configure one or more services. The panoramic chain monitoring in this application embodiment can be considered as one type of service. Taking the vehicle system as an example, the vehicle system can also provide different types of services, such as music playback, intelligent driving control, etc.

[0038] The process involves pre-generating a link development toolkit and obtaining business configuration information. This configuration information can be distributed from a cloud platform or cloud service provider. Developers can configure the business information on the cloud platform, generate the configuration information, and then distribute it to the execution entity. The link development toolkit is parsed, and based on the information defined in the toolkit and the business configuration information, the locations where data points need to be implemented for different application modules are determined and the corresponding data points are implemented. Business processing information is obtained, and based on this information, the application modules to be called, as well as the calling methods and order, are determined. Each application module is then called sequentially.

[0039] S102. In response to the detection that the application interface API for the start of the link in the application module is called, a link identifier is generated. Based on the call of the link node in the link recorded by the tracking point, the node information of the link node is generated based on the link identifier. The node information includes the node identifier, the parent node identifier corresponding to the link node, and the link identifier.

[0040] In this embodiment, the link identifier can be understood as information that uniquely identifies a link trace. For example, the link identifier includes a random ID. Node information can be understood as information describing a link node, such as the unique identifier of the link node, some information about the link node itself, information about the relationship between the link node and other link nodes, etc.; the node identifier can be understood as information that uniquely identifies a link node in this link, and the node identifier of the same link node can be different in different links. The parent node identifier can be understood as information that identifies the parent node of a link node.

[0041] During application module calls, if the link start API, which is embedded in the application module, is detected to be called, link monitoring is initiated and a link identifier is generated. The link identifier can be generated randomly or according to predefined rules. For example, a link identifier generation method or rule can be defined in the link development toolkit, and the link identifier is generated according to this method or rule. For instance, during application module calls, when the corresponding method function in the code is executed, this embodiment pre-embeds embedded points in the code. When the pre-embedded link start API is executed, link monitoring is initiated and a link identifier is generated. During link monitoring, after detecting a call to an embedded link node, this link node is recorded. Based on the link identifier, corresponding node information is generated for the link node, describing the relevant information of the link node and identifying the link node. The node information includes the node identifier, the parent node identifier corresponding to the link node, and the link identifier. That is, when generating the node information of a link node, the node identifier of the link node, the parent node identifier of the link node's corresponding parent node, and the link identifier of the link node's link are determined. Node information is generated based on the node identifier, parent node identifier, and link identifier to reproduce the entire chain based on the node information.

[0042] S103. In response to the detection that the link end API of the embedded point in the application module is called, stop the call to record the link node in the link, and generate link information based on the node information of the recorded link node.

[0043] In this embodiment, link information can be understood as information describing this link, which may include the node identifier of the link node, the execution time and order of the link node, etc. If the link end API of the embedded point in the application module is detected to be called, it is determined to stop link monitoring and stop recording the call of the link node in the link; link information is generated based on the node information of all the link nodes recorded in this link, a link is generated, and the call to the link node is recorded.

[0044] This invention provides a panoramic chain monitoring method for an artificial intelligence application platform, solving problems such as low efficiency and error susceptibility in chain monitoring. By embedding data points in each application module using a chain development toolkit and business configuration information, automated data embedding is achieved, eliminating the need for manual operation, resulting in higher efficiency, higher accuracy, and reduced errors, significantly improving development efficiency. The method acquires business processing information, calls application modules based on this information, and automatically collects chain nodes based on pre-embedded data points. During the collection of chain nodes, when the API for starting a chain point in an application module is called, panoramic chain generation is triggered. First, a chain identifier is generated, providing full-chain connectivity capabilities, linking multiple applications and modules across the front-end and back-end. Then, based on the calls to chain nodes recorded by the data points, node information is generated based on the chain identifier. This node information includes a node identifier, a parent node identifier, and a chain identifier. The node information allows for accurate chain reproduction, enabling effective panoramic chain monitoring of the artificial intelligence application platform.

[0045] Example 2

[0046] Figure 2 This is a flowchart of a panoramic chain monitoring method for an artificial intelligence application platform provided in Embodiment 2 of the present invention. This embodiment is a refinement based on the above embodiments. Figure 2 As shown, the method includes:

[0047] S201. Obtain business processing information and call each application module according to the business processing information. Each application module is a module that has been automatically marked with points according to the link development toolkit and business configuration information.

[0048] Optionally, the method further includes:

[0049] A1. Obtain business configuration information through a distribution channel, which can be either a data push channel or a data pull channel.

[0050] In this embodiment, the distribution channel can be understood as a network channel for transmitting data or information, and can be a wireless communication network channel. In this embodiment, the distribution channel is either a data push channel or a data pull channel. The data push channel is used to actively push data to the execution device, while the data pull channel is used to retrieve data from other devices such as a cloud platform in response to a data pull operation initiated by the execution device.

[0051] The distribution channel for service configuration information can be predetermined. The system can proactively acquire or receive service configuration information from the cloud platform through this channel. Taking the vehicle infotainment system as an example, where the service configuration information is provided by the cloud platform, the vehicle infotainment system can pre-subscribe to the service configuration information. After subscription, the cloud platform can push the service configuration information to the vehicle infotainment system via a data push channel during data updates or periodically. Alternatively, the vehicle infotainment system can proactively acquire the service configuration information from the cloud platform via a data retrieval channel after meeting certain conditions.

[0052] A2. Determine the data entry requirements based on the business configuration information and the link development toolkit, and add data entry points to the code blocks corresponding to each application module where the data entry requirements are met.

[0053] In this embodiment, the tracking requirements can be used to determine whether to perform tracking within a code block. The business configuration information and the link development toolkit are analyzed to determine the function types, business types, and other information requiring tracking, thus obtaining the tracking requirements. For example, two function types, 1 and 2, requiring tracking are predefined in the link development toolkit, and function type 1 is defined in the business configuration information. After comprehensive analysis, it is determined that tracking will be performed on function type 1, with tracking points placed at the beginning and end of all functions corresponding to function type 1. The code blocks corresponding to each application module are scanned to determine locations that meet the tracking requirements, and tracking points are placed at all locations that meet the requirements.

[0054] The instrumentation includes inserting link start API and link end API into the code block corresponding to the application module, inserting node start API at the beginning of the link node, and inserting node end API at the end of the link node.

[0055] The Link Start API and Link End API are used to determine the start and end of link monitoring. The Node Start API and Node End API of a link node are used to determine the link node being called. These APIs are inserted into a code block, and the code block can execute them during execution. When the Link Start API is called, link monitoring begins. If the Node Start API is called to monitor a link node after link monitoring has started, monitoring of that link node will stop after its Node End API is called. Similarly, when the Link End API is called, link monitoring will cease.

[0056] Business configuration information is obtained through a distribution channel, which can be a data push channel or a data pull channel. This information can be acquired proactively or passively to meet diverse business needs. Link start and link end APIs control the start and end of link monitoring, while node start and node end APIs enable monitoring of link nodes, achieving automated, comprehensive link monitoring of the AI ​​application platform.

[0057] For example, Figure 3 An example diagram of a vehicle-mounted system configuration service is provided. This service is used to acquire business configuration information, and the panoramic chain execution monitoring relies on it to control whether data collection is required. The vehicle-mounted system configuration service includes a running platform, a distribution channel, and the vehicle-mounted terminal. On the running platform, business-related configurations are configured, the corresponding configuration group is selected, and finally, the "upload" button is clicked, and a distribution task is selected. The configuration is then distributed to the vehicle-mounted terminal via the distribution channel. The vehicle-mounted terminal receives the distributed configuration and notifies the relevant business party. In this embodiment, the panoramic chain monitoring can act as a business party, enabling monitoring of the panoramic chain.

[0058] Optionally, when the application module is a non-functional component, tracking points are added to the code blocks corresponding to each application module at locations that meet the tracking requirements, including:

[0059] B1. Scan the code blocks of each application module using the link plugin and parse the annotations of the application module's code blocks.

[0060] In this embodiment, the trace plugin can refer to a trace plugin, which is responsible for parsing annotations and performing corresponding code instrumentation based on the annotations. When an application module is not a functional component, the trace plugin scans the code blocks of each application module to determine the annotations of the application module's code blocks and parses the annotations.

[0061] B2. If an annotation that meets the requirements for event tracking is detected, event tracking points will be placed at the start and end positions of the function corresponding to the annotation.

[0062] The annotations are analyzed and matched with the tracking requirements to obtain annotations that meet the requirements. Tracking points are then placed at the start and end positions of the functions corresponding to these annotations. Based on the tracking requirements, it is determined whether this position is the start or end of a link, node, or node. A start API is inserted at the start position, and an end API is inserted at the end position. For the same start position, it can be either a link start position, a node start position, or both. In this case, the link start API should be inserted first, followed by the node start API. Inserting the link start API first ensures timely initiation of link monitoring, while inserting the node start API enables normal monitoring of the link node. Similarly, for the same start position, it can be either a link end position, a node end position, or both. In this case, the node end API should be inserted first, followed by the link end API. Inserting the node end API first ensures normal monitoring of the link node, while inserting the link end API ensures timely cessation of link monitoring. When placing event tracking at the beginning of the function corresponding to the annotation, the link start API and / or node start API are inserted into the code preceding the start position. When placing event tracking at the end position, the link end API and / or node end API are inserted into the code following the explanation position. This avoids data omissions and effectively records the event tracking data. By scanning the code block and analyzing the annotation, event tracking is effectively placed at the start and end positions, achieving effective control over the tracking position and automated event tracking.

[0063] Optionally, when the application module is a functional component, tracking points are placed at locations that meet the tracking requirements within the code blocks corresponding to each application module, including:

[0064] B3. Scan the code blocks of the link adapter through the link plugin and parse the annotations of the code blocks of the link adapter. The code blocks of the link adapter are used to call at least one functional component.

[0065] In this embodiment, the trace adapter can refer to the trace adapter, which is mainly responsible for implementing unified instrumentation for functional components. The code block of the trace adapter is used to call at least one functional component. A functional component can be understood as a kind of plugin that implements different functions, such as navigation function, web search function, etc. An application can have multiple functional components, and a functional component belongs to an application.

[0066] During application module invocation, if the application module is a functional component, it is invoked through a link adapter. Typically, there are hundreds or even thousands of functional components to implement different functions, requiring decoupling of these functions. If instrumentation is implemented in the code block of each functional component, the workload would be substantial. Therefore, this embodiment uses a link adapter to uniformly instrument functional components and invokes them through the link adapter. The link adapter's code block is scanned using a link plugin to determine and parse its annotations.

[0067] B4. If an annotation that meets the requirements for event tracking is detected, event tracking points will be placed at the start and end positions of the function corresponding to the annotation.

[0068] Annotations are analyzed and matched with the tracking requirements to obtain those that meet the requirements. Tracking points are then placed at the start and end positions of the functions corresponding to these annotations. Based on the tracking requirements, it is determined whether this position is the start or end of a link, the start of a node, or the end of a node. A start API is inserted at the start position, and an end API is inserted at the end position. The method for inserting APIs at the start and end positions of links and nodes is described above and will not be repeated here. The tracking method provided in this embodiment avoids data omissions and effectively records the tracking data. By scanning code blocks and analyzing annotations, tracking points are effectively placed at the start and end positions, achieving effective control over tracking positions and automated tracking.

[0069] For example, Figure 4This document provides an example diagram illustrating the interconnected relationship between an AI application platform and the entire value chain. A unified data tracking protocol is used across the upstream and downstream business processes to facilitate subsequent data analysis. The AI ​​main interaction, generative UI, information aggregation framework, TPU, and functional components are all individually integrated with the Trace SDK. The document establishes upstream and downstream relationships between frameworks and their atomic components. Once integrated, newly added functions will automatically reuse this tracking capability, achieving one-time tracking for multiple uses. Key aspects within each framework and atomic capability require separate internal processing. The AI ​​main interaction can be input in at least one of four ways: algorithm, multimodal input, MindGPT context information, and model generation results. During the execution of corresponding operations or data processing based on the AI ​​main interaction, Trace and Span nodes (i.e., link nodes) are created, and DCCs are generated and passed downstream. DCC, an abbreviation for Domain, Command, and Content, identifies the user's dialogue intent and parameters. The business layer executes voice commands based on DCCs; from the moment of voice triggering, all Span nodes in the entire value chain contain this information. Downstream components can call generative UI, information aggregation framework, and TPU. These components can be called directly to obtain the corresponding execution results, or they can be called sequentially for joint data processing. For example, the generative UI can be called first, followed by the TPU or information aggregation framework, and so on. The TPU can also call different functional components across applications. Plugin1, Plugin2, ..., Pluginn are different functional components corresponding to different touch applications. Each functional component belongs to an application, and each application can have one or more functional components to implement different functions.

[0070] S202. In response to the detection that the link start API of the embedded point in the application module is called, a link identifier is generated.

[0071] S203. In response to the detection of the embedded link node, the node start API is called, and the link node is recorded.

[0072] During application module calls, if an API call is detected to begin on a node within the application module that has been embedded, it indicates that this node will be called next, and this call is recorded. Simultaneously, the call process of this node can be recorded, including the data generated and used during this process, or further data analysis can be performed based on the relevant data.

[0073] S204. In response to the detection that the node termination API of the tracked link node is called, the recording of the link node is stopped.

[0074] During application module calls, if the end API of a node in the application module is detected to be called, it indicates that the call to this link node has ended, and recording for this link node stops. By detecting the start and end APIs of the link nodes, the calls to the link nodes are automatically recorded, effectively recording the call process of the link nodes.

[0075] S205. Determine if the link node has a parent node. If it does, execute S206; otherwise, execute S207.

[0076] Analyze the execution order and calling relationships of the link nodes to determine whether a link node has a parent node. If it does, execute S206; otherwise, execute S207.

[0077] S206. Determine the parent node identifier of the link node.

[0078] By determining the node identifier corresponding to the parent node of the link node, the parent node identifier of the link node can be obtained.

[0079] S207. Determine that the parent node identifier of the link node is empty.

[0080] S208. Generate the node identifier of the link node based on the parent node identifier of the link node.

[0081] When generating the node identifier of a link node, the node identifier of the link node is generated based on the node identifier of the link node, and the parent node is reflected in the node identifier of the link node, which makes it convenient to analyze the relationship between link nodes through the node identifier.

[0082] S209. Generate node information of the link node based on the link identifier, node identifier and parent node identifier corresponding to the link node.

[0083] The link identifier, node identifier, and parent node identifier of each link node are written into the corresponding fields to generate the link node's node information. The node information may also include other information about the link node.

[0084] By determining the parent node of a link node, its corresponding parent node identifier is identified, and a node identifier is generated based on the parent node identifier. This establishes a correspondence between the parent node and the node identifier, facilitating subsequent reproduction and restoration of the link. Through effective and accurate link construction and restoration, accurate monitoring of the panoramic link can be achieved, assisting R&D personnel in R&D and analysis.

[0085] Optionally, the node information also includes sequence imprints, and the method further includes:

[0086] Based on the operation logic of the link node and other link nodes in the recorded link, the link relationship between the link node and other recorded link nodes is determined. Based on the link relationship, a sequence imprint of the link node is generated. The sequence imprint is generated in an incremental manner and is used to determine whether the recorded link node is lost.

[0087] The link relationship includes at least one of parallel and serial connections.

[0088] In this embodiment, the sequence mark can be understood as a serial number used to identify the link node. In this embodiment, the sequence mark is generated in an incrementing manner to determine whether the recorded link node is lost. For example, the sequence mark is incremented by 1 in a serial thread. After the link is reproduced, it is determined whether the sequence mark of the link node in the same thread is incremented by 1. If so, it is determined that the link node is not lost; otherwise, it is determined that the link node is lost.

[0089] Analyze the operational logic of the link node and other link nodes in the recorded link to determine the link relationship between the link node and other link nodes in the recorded link. For example, the link relationship between link node 1 and link node 2 is serial, and the link relationship between link node 1 and link node 2 is parallel. Generate the sequence imprint of this link node based on the link relationship and the already generated sequence imprint. For example, the sequence imprint of link node 1 is 1, and the sequence imprint of link node 2 is 2.

[0090] By identifying link nodes through a sequentially increasing sequence, it is possible to determine whether a link node is lost, effectively avoiding the situation where link node loss cannot be detected in time, and promptly detecting data loss.

[0091] Optionally, the link identifier includes: service identifier, link version, and random ID. The link identifier is used to connect the entire link. The service identifier uniquely identifies the service, and the link version identifies the version number to which this link belongs. The link identifier can be transmitted throughout the entire link. In a single-threaded scenario, the vertical flow under the Trace link automatically connects these Span nodes and binds them to a unified link identifier (traceId), and automatically generates a sequence imprint (seqId). In a concurrent scenario, under multi-process and multi-threaded conditions, the service needs to manually associate the upstream and downstream of the Span. During link execution, by transparently transmitting the link identifier, it is clear which link is executing and which node it has reached. Based on these link identifiers, a complete full link is constructed, achieving link coloring. The link identifier can refer to: link identifier, node identifier, sequence imprint, service identifier (BizID), etc.

[0092] S210. In response to the detection that the link end API of the embedded point in the application module is called, stop the call to record the link node in the link, and generate link information based on the node information of the recorded link node.

[0093] For example, Figure 5 This document provides an example diagram of link and data flow. "Link definition" refers to defining a set of link construction rules, which combine multiple logical nodes into a complete logical link. In real-world scenarios, concurrent scenarios involving multiple apps, threads, and processes may occur. To connect the links, a unified identifier is needed for coloring. Nodes in a link typically descend layer by layer from top to bottom, with each node's order starting from 0 and incrementing sequentially. However, in concurrent scenarios, automatic forking is required, with the forked sub-links still starting from 0 and incrementing sequentially. For example... Figure 5 As shown, the monitoring process tracks the execution flow within the business logic. This flow is typically divided into n sub-logic segments, each monitored individually. The relationships between these sub-logic segments are sequential, with the sequence ID automatically incrementing. Each Span can have multiple sub-Span sequences, and the sequence ID of each sub-Span increments from 0. There are four scenarios for generating sub-Span sequences: a. Within the same thread, all Spans generated within the scope of Span A are child nodes of Span A; b. In concurrent scenarios, when constructing a sub-Span, the context information of the parent Span needs to be set to establish the parent-child relationship. The final result is that the internal logic of the business logic is reconstructed into a tree, and the structure of this tree is completely consistent with the written code logic.

[0094] Figure 5 In the code, Span1, Span2...Spann, Span211, Span221, etc., are the node identifiers of the link nodes, and seq is the sequence number of the link nodes. The links are arranged from top to bottom by default, and the sequence automatically increments. Each span has start and end methods. For child spans at the same level, they are sorted by time. Once the parent span executes the end method, the span is considered complete, and the asynchronous child spans can continue to execute according to their own logic.

[0095] For example, the embodiments of this application illustrate the link generation process:

[0096] 1. A user asks about traveling to Beijing through Copilot. A Span1 is built to monitor the code logic in Copilot.

[0097] 2. Copilot outputs the user's question to the information aggregation framework. At this point, the information aggregation framework needs to call other services to complete the answer to the question. Span2 is built to monitor the business logic of the entire information aggregation framework:

[0098] a. There are many logic operations in the same thread within the information aggregation framework. In this case, a sub-Span21 can be generated to monitor the logic in synchronous scenarios.

[0099] b. When dealing with concurrent scenarios, such as retrieving card data, it is necessary to build a Span22 to monitor the logic in asynchronous scenarios;

[0100] c. Monitor all core logic locations until all are fully monitored.

[0101] The core of the panoramic chain execution monitoring and reproduction in this application embodiment lies in the ability to "reproduce on-site". In order to aggregate the scattered data collected by the vehicle terminal into a complete chain, a unified identifier needs to be provided to pass down layer by layer from the starting point of the chain, thereby realizing the coloring of discrete data. Finally, the nodes will be aggregated into a complete chain based on the colored data. The chain is displayed in the form of a tree, and all logic involving concurrent operations will automatically branch.

[0102] Optionally, the method further includes:

[0103] During the call to the link node, the business data collection function set between the node start API and the node end API of the link node is called in sequence, and the business data collected by the business data collection function is recorded.

[0104] Among them, the business data collection function is added between the node start API and node end API of the link node according to the business type.

[0105] In this embodiment, a business data acquisition function can be understood as a function that collects business data. Different business data acquisition functions can collect different types of business data, such as execution time, number of times, etc., to assist developers in research and analysis and facilitate fault location. The business data to be collected is determined based on the business type, and corresponding business data acquisition functions are generated. These functions are added between the start and end APIs of the corresponding link nodes. During the link node call process, different functions are executed sequentially. When a business data acquisition function is called, the business data collected by that function is recorded. Different business data acquisition interfaces can be provided through these functions. Appropriate business data acquisition functions can be inserted according to business needs, flexibly collecting business data to assist developers in research and analysis and facilitate fault location.

[0106] Optionally, the method further includes:

[0107] The reporting conditions are determined based on the link development toolkit;

[0108] In response to the fulfillment of the reporting conditions, the data reporting service is invoked to report the link information, which includes at least one of the following: node information of the link node, data corresponding to the link node, and business data.

[0109] In this embodiment, the reporting conditions can be understood as conditions used to determine whether to report the link information of the monitored panoramic chain. Reporting conditions can be related to information such as time and data volume. The data reporting service can be understood as an application service used for data reporting. Reporting conditions can be predefined in the link development toolkit, and the link development toolkit is parsed to determine the reporting conditions. Real-time detection is performed to check if the reporting conditions are met. If the reporting conditions are met, the data reporting service is invoked to report one or more link information obtained from monitoring. The link information includes at least one of the following: node information of the link node, data corresponding to the link node, and business data. The data corresponding to the link node refers to relevant information generated or used during the link node's invocation process. By setting reporting conditions to report data in a timely manner, the probability of data loss is reduced.

[0110] The reporting conditions in this application embodiment can also be determined through cloud distribution, user settings, or other means.

[0111] The reporting conditions include at least one of the following:

[0112] The time difference between the current time and the previous reporting time exceeds the set time threshold;

[0113] The amount of unreported link information currently exceeds the set data volume threshold.

[0114] In this embodiment, the previous reporting time refers to the time when data was last reported. Time thresholds and data volume thresholds can also be preset based on information such as service type. The current time and the previous reporting time are obtained; for example, the current time is obtained by reading the system clock. The reporting time is saved after each data report, and the latest reporting time is retrieved from the corresponding data storage space as the previous reporting time during subsequent processing. The time difference between the current time and the previous reporting time is calculated, and it is determined whether the time difference exceeds the set time threshold. The data volume of all currently unreported link information is counted; for example, all link information between the current time and the previous reporting time is obtained. This part of the link information is unreported, and it is determined whether the data volume of the currently unreported link information exceeds the set data volume threshold. When at least one of the above conditions is met, it is determined that the reporting condition is met, and the data is reported. By setting reasonable time thresholds for data reporting, it is not necessary to report every time link information is generated, thus avoiding resource waste caused by excessively frequent reporting or data loss caused by reporting too late. Similarly, by setting reasonable data volume thresholds for data reporting, it is not necessary to report every time link information is generated, thus avoiding resource waste caused by excessively frequent reporting.

[0115] For example, Figure 6 A sample diagram of the chain reporting is provided. During the chain execution, each Span node triggers the onStart and onEnd methods (i.e., calls the node start API and node end API). These triggered events are saved to preserve the business context. `addEvent(name, xxx)` and `setAttribute(key, value)` are business data collection functions, collecting different types of business data. Within the Span lifecycle, corresponding business data can be collected by inserting `addEvent(name, xxx)` and `setAttribute(key, value)`. Whether to report data is determined by whether reporting is enabled. If the data is not to be reported, it will not be collected. Within the panoramic chain lifecycle, one or more of the following functions can be implemented: automatic transmission of traceId, automatic calculation of seqId, automatic binding of common parameters, automatic + manual concatenation of parent and child nodes, and automatic parsing of cloud configuration reporting strategies.

[0116] The reporting of the link can be achieved through the data reporting service. After the data collection is completed, the data will be reported to the cloud for data storage through the data reporting service. This service is divided into two reporting methods: real-time reporting and non-real-time reporting. The AI ​​application platform can use the non-real-time reporting method for full-link data collection, which saves data transmission resources. Figure 7An example diagram of a non-real-time data reporting implementation is provided. After data is collected, it is stored in a local data center to form a data file; the data file is processed by data compression, and finally the data is reported.

[0117] This invention provides a panoramic chain monitoring method for an artificial intelligence application platform, solving problems such as low efficiency and error susceptibility in chain monitoring. By scanning code blocks and analyzing annotations, it effectively embeds data at the start and end positions, achieving effective control and automated data embedding. This automated embedding eliminates the need for manual operation, resulting in higher efficiency, accuracy, and reduced errors, significantly improving development efficiency. By determining the parent node of a chain node and generating a node identifier based on it, a correspondence between parent nodes and node identifiers is established, facilitating subsequent chain reproduction and reconstruction. Through effective and accurate chain construction and reconstruction, accurate panoramic chain monitoring can be achieved, assisting in… Research and development personnel conduct research and analysis; by identifying link nodes through a sequentially increasing sequence, it determines whether a link node is lost, effectively avoiding the situation where link node loss cannot be detected in time, and promptly detecting data loss; different business data acquisition interfaces can be provided through business data acquisition functions, and corresponding business data acquisition functions can be inserted according to business needs to flexibly collect business data, assisting research and development personnel in their research and analysis, and facilitating fault location; by setting reporting conditions, data is reported in a timely manner, reducing the probability of data loss; the method provided in this application embodiment realizes effective monitoring of the entire chain of the artificial intelligence application platform, provides the full-link connection capability of the vehicle system's conversational interaction execution, and meets the needs of cross-application call complex scenario connection.

[0118] Example 3

[0119] Figure 8 This is a flowchart of a panoramic chain reconstruction method for an artificial intelligence application platform provided in Embodiment 3 of the present invention. This embodiment is applicable to the situation of panoramic chain reconstruction of an artificial intelligence application platform. The method can be executed by a panoramic chain reconstruction device for the artificial intelligence application platform. The panoramic chain reconstruction device for the artificial intelligence application platform can be implemented in hardware and / or software, and can be configured in a cloud platform; the artificial intelligence application platform includes at least one application module. For example... Figure 8 As shown, the method includes:

[0120] S301. Obtain link information. The link information is generated based on the node information of the link nodes. The node information includes the node identifier, the parent node identifier corresponding to the link node, and the link identifier. The link nodes are recorded based on the instrumentation points of each application module. Each application module automatically performs instrumentation based on the link development toolkit and business configuration information.

[0121] Link information is obtained through wireless communication and other means. The link information is generated based on the node information of the link nodes. The node information includes the node identifier, the parent node identifier corresponding to the link node, and the link identifier.

[0122] S302. Reproduce the link based on the link information.

[0123] The link information is analyzed to determine the node information of the link nodes included in the link information. The node information is analyzed to determine the relationship between the link nodes. The link is constructed based on the relationship between the link nodes and the link is reproduced.

[0124] This invention provides a method for reproducing the entire chain of an artificial intelligence application platform, solving problems such as low efficiency and susceptibility to errors in chain reproduction. By using a chain development toolkit and business configuration information to embed tracking points in each application module, automated tracking is achieved, eliminating the need for manual operation, resulting in higher efficiency, higher accuracy, and reduced errors, significantly improving development efficiency. Based on the calls to chain nodes recorded by the tracking points, node information is generated based on the chain identifier. This node information includes the node identifier, parent node identifier, and chain identifier. The node information allows for accurate chain reproduction, enabling effective monitoring of the entire chain of the artificial intelligence application platform.

[0125] Optionally, business configuration information can be sent by the cloud platform through a distribution channel, which can be a data push channel or a data pull channel.

[0126] Optionally, the link can be reproduced based on the link information, including:

[0127] C1. Determine the relationship between each link node based on the node identifier and the parent node identifier. The relationship between each link node can be either parallel or serial.

[0128] Analyze the node identifier and parent node identifier to determine the relationship between link nodes. For example, check if a link node has a parent node. If two link nodes have the same parent node identifier, they are determined to have the same parent node and are in a parallel relationship. If a link node does not have a parent node, the relationship between the link node and its preceding executed link node is serial.

[0129] C2. Reproduce the links based on the relationships between the links and generate the links.

[0130] The links are reproduced based on the relationships between the links, the links that are executed sequentially and the links that are executed in parallel are determined, and the links are constructed to achieve the reproduction of the links.

[0131] C3. Identify the generated links based on the link identifier.

[0132] The generated links are identified based on link identifiers to achieve unique identification of each link. The relationships between link nodes are analyzed using node identifiers and parent node identifiers to reproduce the links, enabling fast and accurate link reproduction. The unique identification of each link facilitates subsequent link analysis.

[0133] Optionally, the node information also includes a sequence imprint, which is generated in a sequentially increasing manner; the method also includes: determining whether the reproduced link has lost a link node based on the sequence imprint.

[0134] The sequence imprints of link nodes in the path are analyzed to determine if they increment according to a set pattern. For link nodes within the same thread, their sequence imprints increment sequentially. Within the scope of a single link node, the link nodes within that scope are its child nodes. The relationships between child nodes can be parallel or sequential. For example, if child node 1 and child node 2 are sequential, and child node 1 and child node 3 are parallel, then child node 1's sequence imprint is 0, child node 2's sequence imprint is 1, indicating the sequential relationship of the link nodes in this path, and child node 3's sequence imprint is 0, indicating that the path containing child node 3 is parallel to the paths containing child node 1 and child node 2. By identifying link nodes through sequentially incrementing sequence imprints, the system can determine if a link node is lost, effectively preventing the failure to detect link node losses in a timely manner and ensuring prompt detection of data loss.

[0135] For example, Figure 9 This paper presents a system architecture diagram for panoramic chain execution monitoring and reproduction. The system consists of two subsystems: a panoramic chain data acquisition system and a cloud storage / aggregation and display system. The panoramic chain data acquisition system builds a complete panoramic chain tracking toolchain on top of the system services for use by the vehicle's infotainment system. The cloud storage / aggregation and display system provides data storage capabilities for the vehicle's infotainment system, as well as data aggregation, display, and alert functions. The panoramic chain data acquisition system includes the following: the panoramic chain tracking toolchain comprises three tools: Trace Plugin, Trace Adapter, and Trace UI Tool; it also includes the Trace SDK, vehicle infotainment system configuration service, and data reporting service.

[0136] Trace SDK: Responsible for formulating the unified data protocol for the entire chain, implementing the core of connecting upstream and downstream links, collecting data, and providing a flexible and unified API for business users to call.

[0137] Trace Plugin: Responsible for parsing annotations and performing corresponding code instrumentation based on the annotations.

[0138] Trace Adapter: This is the trace adapter, which is mainly responsible for implementing unified instrumentation for functional components.

[0139] Trace UI Tool: Provides a visual trace reconstruction tool to verify whether the trace structure and nodes are correct. It is mainly used when analyzing a specific trace.

[0140] The vehicle infotainment system configuration service is used for configuration updates and retrieval, such as obtaining business configuration information from the cloud storage / aggregation and display system and performing configuration updates. The Trace SDK reads the configuration from the vehicle infotainment system configuration service and implements automatic data tracking based on the panoramic chain tracking toolchain. After tracking, data is collected and reported to the cloud storage / aggregation and display system. The cloud storage / aggregation and display system stores the received data and performs corresponding aggregation and display.

[0141] in, Figure 9 In this context, "Reporter" refers to the data reporter; the same data can be reported to different backends through different Reporters. "Logger" is the end-to-end internal logging tool that controls log printing in different environments. "Trace" represents a complete chain, spanning the entire business logic from top to bottom. "Span" represents a span node in the chain, monitoring the execution of a certain piece of code logic. "Event" represents an event triggered at a certain time within the "Span".

[0142] Optionally, the method also includes: displaying the link information in the form of a chart.

[0143] Taking the panoramic chain monitoring implemented by the vehicle's infotainment system and the panoramic chain reconstruction implemented by the cloud as an example, the vehicle's infotainment system uses a full-link data acquisition SDK to transmit data from the AI ​​application platform to the cloud storage system. Subsequently, this data undergoes a refined processing flow and is ultimately integrated and presented in an intuitive chart format. Utilizing these detailed data resources, existing problems can be accurately identified and promptly fed back to the AI ​​application platform for remedial measures. Furthermore, this data also becomes valuable BI (Business Intelligence) decision analysis material, providing strong support for in-depth insights and strategic formulation. In any embodiment of this application, "cloud" can refer to a cloud platform, and a cloud platform can refer to the cloud itself.

[0144] For example, Figure 10This document provides an example diagram illustrating data acquisition and processing. The AI ​​application platform integrates with the Panoramic Chain data acquisition SDK to collect data. The collected data is then reported to the cloud via a data reporting service. The cloud stores the data and, through a data synchronization service, stores it in a database. This database stores the raw data. The data synchronization service can also synchronize data to other databases for further storage and business analysis. The raw data can undergo one or more processing steps, such as correlation fusion, extraction, and other methods. The processed data is stored in data assets for analysis, which can be used for BI decision analysis and relational computation. BI stands for Business Intelligence, a set of concepts and methods used to collect, organize, analyze, and present information so that enterprises or organizations can better understand their business situation and make decisions accordingly. Data can be synchronized to BI decision analysis to assist in its implementation, or data can be queried during BI decision analysis, or data queries can be used to obtain data provided by BI decision analysis, etc. Data obtained from relational computation can be displayed in the form of charts, etc. The data relationships obtained after relational computation can be stored in graph storage. For the data in graph storage, further relational extraction can be performed to extract primary indicators, secondary indicators, etc. For example, first analyze the primary indicators to perform relational extraction, and then further extract the secondary indicators from the data already statistically analyzed from the primary indicators. Multi-level indicators can be extracted through iterative extraction. The extracted data is stored in data assets. The raw data and the data in graph storage can be analyzed to perform root cause reasoning problem repair, identify existing problems, and improve the AI ​​application platform.

[0145] This invention provides a method for reproducing the entire chain of an artificial intelligence application platform, solving problems such as low efficiency and susceptibility to errors in chain reproduction. By using a chain development toolkit and business configuration information to embed tracking points in each application module, automated tracking is achieved, eliminating the need for manual operation, resulting in higher efficiency, higher accuracy, and reduced errors, significantly improving development efficiency. Node information allows for accurate chain reproduction, enabling effective monitoring of the entire chain of the artificial intelligence application platform. Cloud-based aggregation and display capabilities and chain reconstruction capabilities enhance problem localization and analysis efficiency; one-click restoration of the entire chain execution process is supported, and rich chain call display formats are provided, reducing the resource investment required for troubleshooting business issues.

[0146] Example 4

[0147] Figure 11 This is a schematic diagram of a panoramic chain monitoring device for an artificial intelligence application platform provided in Embodiment 4 of the present invention. The artificial intelligence application platform includes at least one application module, such as... Figure 11 As shown, the device includes: a service information acquisition module 31, a node information acquisition module 32, and a link information generation module 33.

[0148] The business information acquisition module 31 is used to acquire business processing information and call each of the application modules according to the business processing information, wherein each of the application modules is a module that has been automatically embedded according to the link development toolkit and business configuration information;

[0149] The node information acquisition module 32 is used to respond to the call of the application programming interface API for the start of the link in the application module when the embedded point is detected, generate a link identifier, generate node information of the link node according to the link node in the embedded point record, and generate node information of the link node according to the link identifier. The node information includes a node identifier, a parent node identifier corresponding to the link node, and a link identifier.

[0150] The link information generation module 33 is used to stop recording the link node calls in the link in response to the detection that the link end API of the embedded point in the application module is called, and to generate link information based on the node information of the recorded link node.

[0151] This invention provides a panoramic chain monitoring device for an artificial intelligence application platform, solving problems such as low efficiency and error susceptibility in chain monitoring. By embedding data points in each application module using a chain development toolkit and business configuration information, automated data embedding is achieved, eliminating the need for manual operation, resulting in higher efficiency, higher accuracy, and reduced errors, significantly improving development efficiency. The device acquires business processing information, calls application modules based on this information, and automatically collects chain nodes based on pre-embedded data points. During the collection of chain nodes, when the API for starting a chain point in an application module is called, panoramic chain generation is triggered. First, a chain identifier is generated, providing full-chain connectivity capabilities through a unified identifier, enabling the connection of multiple applications and modules across the front-end and back-end. Then, based on the calls to chain nodes recorded by the data points, node information is generated based on the chain identifier. This node information includes a node identifier, a parent node identifier, and a chain identifier. The node information allows for accurate chain reconstruction, enabling effective panoramic chain monitoring of the artificial intelligence application platform.

[0152] Optionally, the device may also include:

[0153] The configuration information acquisition module is used to acquire business configuration information through a distribution channel, which is either a data push channel or a data pull channel.

[0154] The event tracking module is used to determine the event tracking requirements based on the business configuration information and the link development toolkit, and to track the locations in the code blocks corresponding to each application module that meet the event tracking requirements.

[0155] The event tracking includes inserting a link start API and a link end API into the code block corresponding to the application module, inserting a node start API at the beginning of a link node, and inserting a node end API at the end of a link node.

[0156] Optionally, when the application module is a non-functional component, the data tracking module includes:

[0157] The first scanning unit is used to scan the code blocks of each application module through the link plugin and parse the annotations of the code blocks of the application modules.

[0158] The first data entry unit is used to perform data entry at the start and end positions of the function corresponding to the annotation if an annotation that meets the data entry requirements is detected.

[0159] Optionally, when the application module is a functional component, the data tracking module includes:

[0160] The second scanning unit is used to scan the code block of the link adapter through the link plugin and parse the annotation of the code block of the link adapter, wherein the code block of the link adapter is used to call at least one of the functional components;

[0161] The second tracking unit is used to track the start and end positions of the function corresponding to the annotation if an annotation that meets the tracking requirements is detected.

[0162] Optionally, the node information acquisition module 32 includes:

[0163] The node call recording unit is used to record the call of the link node in response to the detection that the node start API of the embedded link node is called.

[0164] The node call recording stop unit is used to stop recording the link node in response to the node end API being called when the node of the embedded link node is detected.

[0165] Optionally, the node information acquisition module 32 includes:

[0166] The parent node determination unit is used to determine whether the link node has a parent node. If it does, it determines the parent node identifier of the link node; if it does not, it determines that the parent node identifier of the link node is empty.

[0167] The identifier generation unit is used to generate a node identifier for the link node based on the parent node identifier of the link node;

[0168] The node information generation unit is used to generate node information of the link node based on the link identifier, the node identifier and the parent node identifier corresponding to the link node.

[0169] Optionally, the node information further includes a sequence imprint, and the device further includes:

[0170] The sequence imprint generation module is used to determine the link relationship between the link node and other recorded link nodes according to the operation logic of the link node and other recorded link nodes, and generate a sequence imprint of the link node according to the link relationship. The sequence imprint is generated in a sequentially increasing manner, and the sequence imprint is used to determine whether the recorded link node is lost.

[0171] The link relationship includes at least one of parallel and serial connections.

[0172] Optionally, the device may also include:

[0173] The business data acquisition module is used to sequentially call the business data acquisition function set between the node start API and the node end API of the link node during the calling process of the link node, and record the business data collected by the business data acquisition function.

[0174] The business data acquisition function is added between the node start API and node end API of the link node according to the business type.

[0175] Optionally, the device may also include:

[0176] The reporting condition determination module is used to determine the reporting conditions based on the link development toolkit.

[0177] The reporting condition generation module is used to call the data reporting service to report the link information in response to the fulfillment of the reporting conditions. The link information includes at least one of the following: node information of the link node, data corresponding to the link node, and business data.

[0178] The reporting conditions include at least one of the following:

[0179] The time difference between the current time and the previous reporting time exceeds the set time threshold;

[0180] The amount of unreported link information currently exceeds the set data volume threshold.

[0181] The panoramic chain monitoring device of the artificial intelligence application platform provided in the embodiments of the present invention can execute the panoramic chain monitoring method of the artificial intelligence application platform provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0182] Example 5

[0183] Figure 12 This is a schematic diagram of a panoramic chain reproduction device for an artificial intelligence application platform provided in Embodiment 5 of the present invention. The artificial intelligence application platform includes at least one application module, such as... Figure 12 As shown, the device includes a link information acquisition module 41 and a link reproduction module 42.

[0184] Link information acquisition module 41 is used to acquire link information. The link information is generated based on the node information of the link node. The node information includes a node identifier, a parent node identifier corresponding to the link node, and a link identifier. The link node is recorded based on the data points of each application module. Each application module automatically performs data point recording based on the link development toolkit and business configuration information.

[0185] The link reproduction module 42 is used to reproduce the link based on the link information.

[0186] This invention provides a panoramic chain reproduction device for an artificial intelligence application platform, solving problems such as low efficiency and error susceptibility in chain reproduction. By embedding data points in each application module using a chain development toolkit and business configuration information, automated data embedding is achieved, eliminating the need for manual operation, resulting in higher efficiency, higher accuracy, and reduced errors, significantly improving R&D efficiency. Based on the calls to chain nodes recorded by the data embedding, node information is generated based on the chain identifier. This node information includes a node identifier, a parent node identifier, and a chain identifier. The node information allows for accurate chain reproduction, enabling effective panoramic chain monitoring of the artificial intelligence application platform.

[0187] Optionally, the link reproduction module 42 is specifically used for: determining the relationship between each link node according to the node identifier and the parent node identifier, wherein the relationship between each link node is parallel or serial; reproducing the link according to the relationship between each link node to generate a link; and identifying the generated link according to the link identifier.

[0188] Optionally, the node information further includes a sequence imprint, which is generated in a sequentially increasing manner; the device further includes:

[0189] The node loss determination module is used to determine whether the reproduced link has lost a link node based on the sequence imprint.

[0190] The panoramic chain reproduction device of the artificial intelligence application platform provided in the embodiments of the present invention can execute the panoramic chain reproduction method of the artificial intelligence application platform provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0191] Example 6

[0192] Figure 13 This is a schematic diagram of an electronic device provided for an embodiment of the present invention. The electronic device can be used to implement the methods provided in any embodiment of the present invention. 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 assistants, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein. The electronic device in this embodiment can be deployed in a vehicle or on a cloud platform.

[0193] like Figure 13 As shown, the electronic device 50 includes at least one processor 51 and a memory, such as a read-only memory (ROM) 52 and a random access memory (RAM) 53, communicatively connected to the at least one processor 51. The memory stores computer programs executable by the at least one processor. The processor 51 can perform various appropriate actions and processes based on the computer program stored in the ROM 52 or loaded into the RAM 53 from storage unit 58. The RAM 53 can also store various programs and data required for the operation of the electronic device 50. The processor 51, ROM 52, and RAM 53 are interconnected via a bus 54. An input / output (I / O) interface 55 is also connected to the bus 54.

[0194] Multiple components in electronic device 50 are connected to I / O interface 55, including: input unit 56, such as keyboard, mouse, etc.; output unit 57, such as various types of monitors, speakers, etc.; storage unit 58, such as disk, optical disk, etc.; and communication unit 59, such as network card, modem, wireless transceiver, etc. Communication unit 59 allows electronic device 50 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0195] Processor 51 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 51 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 51 performs the various methods and processes described above, such as panoramic chain monitoring or reproduction methods in an artificial intelligence application platform.

[0196] In some embodiments, the panoramic chain monitoring or reproduction method of the artificial intelligence application platform can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 58. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 50 via ROM 52 and / or communication unit 59. When the computer program is loaded into RAM 53 and executed by processor 51, one or more steps of the panoramic chain monitoring or reproduction method of the artificial intelligence application platform described above can be performed. Alternatively, in other embodiments, processor 51 can be configured to perform the panoramic chain monitoring or reproduction method of the artificial intelligence application platform by any other suitable means (e.g., by means of firmware).

[0197] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0198] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0199] This invention provides a computer program product, which includes a computer program that, when executed by a processor, implements the panoramic chain monitoring or reproduction method of the artificial intelligence application platform described in any embodiment of this invention.

[0200] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0201] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, 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 sound input, voice input, or tactile input).

[0202] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0203] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. 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 cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0204] 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 invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0205] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. 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 invention should be included within the scope of protection of this invention.

Claims

1. A panoramic chain monitoring method for an artificial intelligence application platform, characterized in that, The artificial intelligence application platform includes at least one application module, and the method includes: Obtain business processing information, and call each of the application modules according to the business processing information, wherein each of the application modules is a module that has been automatically embedded according to the link development toolkit and business configuration information; In response to the detection that the link start application programming interface (API) of the embedded point in the application module is called, a link identifier is generated. According to the call of the link node in the link recorded by the embedded point, the node information of the link node is generated according to the link identifier. The node information includes the node identifier, the parent node identifier corresponding to the link node, and the link identifier. In response to the detection that the link end API of the embedded point in the application module is called, the recording of the link node calls in the link is stopped, and link information is generated based on the node information of the recorded link node.

2. The method according to claim 1, characterized in that, Also includes: Business configuration information is obtained through a distribution channel, which is either a data push channel or a data pull channel. Based on the business configuration information and the link development toolkit, determine the tracking requirements, and track the locations in the code blocks corresponding to each application module that meet the tracking requirements. The event tracking includes inserting a link start API and a link end API into the code block corresponding to the application module, inserting a node start API at the beginning of a link node, and inserting a node end API at the end of a link node.

3. The method according to claim 2, characterized in that, When the application module is a non-functional component, the step of embedding points at locations in the code blocks corresponding to each application module that meet the embedding requirements includes: The code blocks of each application module are scanned using the link plugin, and the annotations of the code blocks of the application modules are parsed. If an annotation that meets the requirements for embedding is detected, embedding points are performed at the start and end positions of the function corresponding to the annotation.

4. The method according to claim 2, characterized in that, When the application module is a functional component, the step of embedding points at locations in the code blocks corresponding to each application module that meet the embedding requirements includes: The code blocks of the link adapter are scanned by the link plugin, and the annotations of the code blocks of the link adapter are parsed, wherein the code blocks of the link adapter are used to call at least one of the functional components; If an annotation that meets the requirements for embedding is detected, embedding points are performed at the start and end positions of the function corresponding to the annotation.

5. The method according to claim 1, characterized in that, The invocation of link nodes in the link based on the embedded point record includes: In response to the detection of a tracked link node, the node start API is called, and the link node is recorded as called. In response to the detection that the node termination API of the tracked node is called, recording of the tracked node is stopped.

6. The method according to claim 1, characterized in that, The step of generating the node information of the link node based on the link identifier includes: Determine whether the link node has a parent node. If it does, determine the parent node identifier of the link node. If it does not, determine that the parent node identifier of the link node is empty. The node identifier of the link node is generated based on the parent node identifier of the link node; The node information of the link node is generated based on the link identifier, the node identifier, and the parent node identifier corresponding to the link node.

7. The method according to any one of claims 1-6, characterized in that, The node information also includes sequence imprints, and the method further includes: Based on the operation logic of the link node and other link nodes in the recorded link, the link relationship between the link node and other recorded link nodes is determined, and a sequence imprint of the link node is generated based on the link relationship. The sequence imprint is generated in a sequentially increasing manner and is used to determine whether the recorded link node is lost. The link relationship includes at least one of parallel and serial connections.

8. The method according to any one of claims 1-7, characterized in that, Also includes: During the invocation of the link node, the business data acquisition function set between the node start API and the node end API of the link node is invoked in sequence, and the business data collected by the business data acquisition function is recorded. The business data acquisition function is added between the node start API and node end API of the link node according to the business type.

9. The method according to any one of claims 1-8, characterized in that, Also includes: The reporting conditions are determined based on the aforementioned link development toolkit; In response to the fulfillment of the reporting conditions, the data reporting service is invoked to report the link information, which includes at least one of the following: node information of the link node, data corresponding to the link node, and business data. The reporting conditions include at least one of the following: The time difference between the current time and the previous reporting time exceeds the set time threshold; The amount of unreported link information currently exceeds the set data volume threshold.

10. A method for reproducing the panoramic chain of an artificial intelligence application platform, characterized in that, The artificial intelligence application platform includes at least one application module, and the method includes: Obtain link information, which is generated based on the node information of the link nodes. The node information includes a node identifier, the parent node identifier corresponding to the link node, and a link identifier. The link nodes are recorded based on the data points of each application module. Each application module automatically performs data point recording based on the link development toolkit and business configuration information. The link is reproduced based on the link information.

11. The method according to claim 10, characterized in that, The step of reproducing the link based on the link information includes: The relationship between each link node is determined based on the node identifier and the parent node identifier, and the relationship between each link node is either parallel or serial. The links are reproduced and generated based on the relationships between the links. The generated links are identified based on the link identifier.

12. The method according to any one of claims 10-11, characterized in that, The node information also includes a sequence imprint, which is generated in a sequentially increasing manner; the method further includes: The sequence print is used to determine whether the reproduced link has lost a link node.

13. An electronic device, characterized in that, The electronic device includes: At least one processor, and a memory communicatively connected to said at least one processor; The memory stores a computer program that can be executed by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to execute the panoramic chain monitoring method of the artificial intelligence application platform according to any one of claims 1-9 or the panoramic chain reproduction method of the artificial intelligence application platform according to any one of claims 10-12.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the panoramic chain reproduction method of the artificial intelligence application platform according to any one of claims 1-9 or the panoramic chain reproduction method of the artificial intelligence application platform according to any one of claims 10-12.

15. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the panoramic chain monitoring method of the artificial intelligence application platform according to any one of claims 1-9 or the panoramic chain reproduction method of the artificial intelligence application platform according to any one of claims 10-12.

Images