Supply chain low-code platform service management method, equipment and medium

Abstract

The invention discloses a supply chain low-code platform service management method and device and a medium, and the method comprises the steps: receiving and analyzing Java method semantic information in a Jar package file to generate a standardized service number, or analyzing a protocol type and a parameter format of external API address information, and extracting service attribute data to generate service operation recommendation data; generating service registration information after confirmation of the developer, and storing the service registration information in a database to construct a service portrait; in response to the service calling request, executing data format conversion and protocol adaptation operation of calling parameters based on a machine learning model; acquiring a service calling log, identifying an abnormal calling event based on a preset abnormal detection model, and pushing abnormal early warning information to a developer; and constructing a user portrait based on developer information, matching and recommending an adaptive service set for the developer based on the user portrait and the service portrait, and allowing the developer to position a target service through semantic search.

Description

Technical Field

[0001] This application relates to the fields of low-code development and supply chain technology, and in particular to a method, device and medium for managing supply chain low-code platform services. Background Technology

[0002] As enterprises deepen their digital transformation of supply chains, low-code development platforms have become key tools due to their ability to rapidly build and iterate business systems. Service management, as a core component of such platforms, is responsible for the unified registration, management, and scheduling of reusable native Java methods and external integration APIs within the platform. Its efficiency and intelligence directly determine the development speed and operational stability of supply chain business applications. However, existing low-code platform service management solutions still heavily rely on manual operations and experience-based judgment by developers, exhibiting significant shortcomings in efficiency, adaptability, and reliability.

[0003] Current service registration processes typically require developers to manually parse Java method signatures in JAR files or consult external API documentation, and then fill in attribute information such as service number, parameter types, and business function descriptions one by one. When faced with complex JAR files containing numerous methods or external interfaces with cumbersome parameter structures, this process is not only time-consuming and lengthy, but also highly susceptible to human error leading to incorrect or inconsistent attribute annotations, creating potential problems for subsequent service calls. During the service call phase, due to inherent differences in calling protocols and data formats between local methods and external APIs, developers must write additional adaptation code for conversion. This not only increases the development burden but also makes it difficult to guarantee a high success rate for cross-service calls due to the complexity of the adaptation logic.

[0004] Furthermore, existing solutions generally lag behind in monitoring service operation status and lack the ability to analyze call logs and identify anomalies in real time. Operations personnel often have to manually trace logs to locate problems after a business system fails, resulting in excessively long average recovery times and failing to meet the high continuity requirements of supply chain operations. At the same time, the platform's large pool of service resources lacks an effective intelligent recommendation mechanism, requiring both new and experienced developers to spend significant time familiarizing themselves with and selecting services, reducing the platform's usability and service reusability. Summary of the Invention

[0005] This application provides a supply chain low-code platform service management method, device, and medium to solve the above-mentioned technical problems.

[0006] On one hand, embodiments of this application provide a supply chain low-code platform service management method, including: Receive business JAR package files uploaded by developers or input external API address information, and parse the Java method semantic information in the JAR package file to generate a standardized service number, or parse the protocol type and parameter format of the external API address information to extract service attribute data to generate business operation recommendation data; After the developer confirms the information, complete service registration information is generated and stored in the database to build a service profile. In response to the developer's service call request, the machine learning model trained based on historical call data automatically performs data format conversion and protocol adaptation operations for the call parameters; Collect service call logs in real time and identify abnormal call events based on a preset anomaly detection model to push anomaly warning information to the corresponding developers; User profiles are built based on developer information. Based on the user profiles and service profiles, a recommendation algorithm is used to match and recommend a set of suitable services to the developers, and the developers are allowed to locate target services through semantic search.

[0007] In one implementation of this application, parsing the Java method semantic information in the Jar package file to generate a standardized service number specifically includes: Load the business JAR file, and traverse and parse the Java class and method definitions in the business JAR file; Using a pre-trained language model based on the Transformer architecture, the parsed Java method source code is encoded to generate high-dimensional semantic vectors; Analyze the high-dimensional semantic vector to automatically identify and extract the method name, the parameter types that constitute the method input, and the return value type that constitutes the method output, which represent the function of the method. Based on predefined number generation rules, the extracted attribute data is concatenated to generate a standardized service number string.

[0008] In one implementation of this application, parsing the protocol type and parameter format of the external API address information and extracting service attribute data to generate business operation recommendation data specifically includes: Send a probe request to the external API address to obtain the interface description information of the external API address; The interface description information is parsed to automatically identify the protocol type corresponding to the network request protocol used by the external API address; the protocol type includes HTTP or HTTPS. The request body and response body structure definitions in the interface description information are parsed to automatically identify the parameter format of the external API address, wherein the parameter format includes JSON or XML; The identified protocol types and parameter formats are mapped to predefined service source categories, and the service source attributes are matched and labeled as external integration services.

[0009] In one implementation of this application, a machine learning model trained based on historical call data automatically performs data format conversion and protocol adaptation operations for the call parameters, specifically including: Collect historical service call records and use these records as training samples. Train a random forest machine learning model with the source parameter format as input and the target parameter format as output. The historical service call records contain the source parameter format, the target parameter format, and the mapping relationship of successful conversion. When a new service call request is triggered, the source parameter format of the current call is input into the trained random forest model, and the predicted target parameter format is output. Based on the predicted target parameter format, serialization or deserialization operations are performed on the parameter object to achieve data format conversion and protocol adaptation.

[0010] In one implementation of this application, service call logs are collected in real time, and abnormal call events are identified based on a preset anomaly detection model, specifically including: Continuously collect service call response time data to form a time series dataset; The time series dataset is trained using a Long Short-Term Memory (LSTM) network model to learn normal fluctuation patterns in service calls in order to construct a service operation status baseline; the service operation status baseline includes a predefined normal range of response time within a specific time window; The call response time of the real-time monitoring service is compared with the service operation status baseline. When the call response time continuously exceeds the preset baseline threshold range, the current call event is marked as an abnormal call event. Extract the multidimensional feature data when the abnormal call event occurs, and input the multidimensional feature data into a pre-trained gradient boosting decision tree classification model for inference, and output the fault type corresponding to the abnormal call event.

[0011] In one implementation of this application, the developer is allowed to locate the target service through semantic search, specifically including: Receive natural language query statements input by developers, and perform word segmentation and stop word removal on the natural language query statements to obtain a keyword set; Calculate the term frequency inverse document frequency (TNV) values ​​of the keyword set and all service function description texts in the service profile, and construct the text vectors of the corresponding keywords based on the TNV values; Calculate the cosine similarity between the text vector corresponding to each keyword in the keyword set and the service description vector corresponding to the service function description text, and sort the services in the service profile according to the cosine similarity, so as to return and locate the service with the highest similarity in the sorted list as the target service.

[0012] In one implementation of this application, before receiving the business JAR file uploaded by the developer or the external API address information entered, the method further includes: The service management system is deployed on the application server of the supply chain low-code platform, and a database instance is integrated to persist storage service metadata and call historical records and model training results. Import scenario tag data and historical service data from the supply chain business knowledge base to form a training dataset; the scenario tag data includes semantic descriptions of expense reimbursement management, logistics traceability and order fulfillment, and the historical service data includes service numbers, parameter types, call frequencies and error logs of registered services; Based on the training dataset, natural language processing techniques are used to train a semantic parsing model, a service recommendation model, and an anomaly detection model, and the parameters of the semantic parsing model, the service recommendation model, and the anomaly detection model are initialized.

[0013] In one implementation of this application, it further includes: Feedback data from developers is collected periodically through the user interface at preset time intervals, and the feedback data is integrated with the corresponding original service data and call data to form a new training sample set; the feedback data includes recommendation accuracy score data, warning timeliness comment data, and service usage problem description data; The semantic parsing model, service recommendation model, and anomaly detection model are incrementally trained periodically using new training sample sets, and the optimized models are redeployed to the production environment.

[0014] On the other hand, embodiments of this application also provide a supply chain low-code platform service management device, the device comprising: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are then executed by the at least one processor to enable the at least one processor to perform a supply chain low-code platform service management method as described above.

[0015] On the other hand, embodiments of this application also provide a non-volatile computer storage medium storing computer-executable instructions, which, when executed, implement a supply chain low-code platform service management method as described above.

[0016] This application provides a supply chain low-code platform service management method, device, and medium, which has at least the following beneficial effects: By automatically extracting key attributes such as method name, parameter type, and return value type through an AI semantic parsing engine and intelligently generating standardized service numbers, the workload of developers is significantly reduced, and the error-prone nature of manual parsing is solved, ensuring the accuracy and consistency of service attributes. A machine learning model trained on historical call data enables automatic conversion between parameter formats and calling protocols. Real-time monitoring and anomaly warning mechanisms proactively and promptly detect potential faults such as response timeouts and parameter mismatches, ensuring the continuity and stability of the supply chain business system. By constructing user and service profiles and utilizing intelligent recommendation algorithms and semantic search technology, the system can accurately recommend or locate the services needed by developers based on their context and natural language descriptions, accelerating the development process of business applications and improving the development experience and service reusability of the platform. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a supply chain low-code platform service management method provided in this application embodiment; Figure 2 This is a schematic diagram of the internal structure of a supply chain low-code platform service management device provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0020] Figure 1This is a flowchart illustrating a supply chain low-code platform service management method provided in an embodiment of this application.

[0021] The analysis method involved in the embodiments of this application can be implemented by a terminal device or a server, and this application does not impose any special limitations on it. For ease of understanding and description, the following embodiments are all described in detail using a server as an example.

[0022] It should be noted that the server can be a single device or a system composed of multiple devices, i.e., a distributed server. This application does not make any specific limitations on this.

[0023] like Figure 1 As shown in the embodiment of this application, a supply chain low-code platform service management method includes: Step 101: Receive the business JAR package file uploaded by the developer or the external API address information input, and parse the Java method semantic information in the JAR package file to generate a standardized service number, or parse the protocol type and parameter format of the external API address information to extract service attribute data to generate business operation recommendation data.

[0024] In this embodiment, the system is first deployed on the application server of the supply chain low-code platform (compatible with Docker containerized deployment), integrating MySQL database storage service attributes and call logs. For example, the service management system can be packaged as a standard Docker image and deployed on the platform's Kubernetes cluster or a traditional application server (such as Tomcat), thereby ensuring its collaborative operation with other modules of the low-code platform.

[0025] Secondly, scenario tag data and historical service data need to be imported from the supply chain business knowledge base to form a training dataset. The supply chain business knowledge base is a pre-built structured or semi-structured database containing knowledge of the supply chain domain. Scenario tag data mainly includes standardized semantic descriptions of typical business scenarios such as expense reimbursement management, logistics traceability, and order fulfillment. For example, the "expense reimbursement management" scenario might be associated with keywords such as "invoice," "approval," and "amount." Historical service data refers to service call information that may exist in older versions of systems or related systems and has been organized, including the service number, parameter type, call frequency, and error logs of registered services. Understandably, importing and integrating this data into a unified "training dataset" is a necessary step to provide learning material for the AI ​​model, enabling it to understand the specific context and historical patterns of supply chain business.

[0026] Finally, based on the prepared training dataset, the system performs the core model training and initialization tasks. Specifically, natural language processing techniques are used to train the semantic parsing model, service recommendation model, and anomaly detection model. For example, for the semantic parsing model (such as a BERT-based model), it is trained using data containing a large amount of Java method code and its functional descriptions, enabling it to learn to accurately extract semantic attributes from method signatures and comments. For the service recommendation model (such as a collaborative filtering model), it is trained using historical data containing user-service interaction records (such as which developers called which services) to discover potential recommendation patterns. For the anomaly detection model (such as an LSTM time-series model), it is trained using normal service call time-series data to learn a baseline of normal behavior, and its classifier (such as XGBoost) is trained using data containing anomaly labels. It should be noted that after training, the system initializes the parameters of these models and saves them to a specified location so that they can be quickly loaded and invoked in subsequent steps.

[0027] In this embodiment, data reception is achieved through a visual interactive interface provided by the supply chain low-code platform. For example, developers can upload local business JAR files by dragging and dropping or clicking buttons, or fill in known external API Uniform Resource Locators in the specified input box.

[0028] Understandably, business JAR files typically contain one or more Java classes and methods that encapsulate specific business logic, such as invoice verification and inventory queries in a supply chain scenario. External API address information, on the other hand, points to service interfaces outside the platform that can be accessed over the network, such as third-party logistics tracking APIs.

[0029] Specifically, for processing Jar package files, the system first loads and parses the business Jar package file, traversing the Java class and method definitions within it. For example, the system uses a pre-trained language model based on the Transformer architecture (such as BERT) to encode the parsed Java method source code, generating a high-dimensional semantic vector. By analyzing this semantic vector, the system can automatically identify and extract deep semantic information such as the method name representing the method's function, the types of parameters constituting the method's input, and the type of the method's return value. Finally, based on predefined numbering rules, such as a hierarchical structure of "product.module.application.entity.method name," the extracted attribute data is concatenated to generate a standardized service number string.

[0030] It's important to note that when processing external API address information, the system sends a probe request to the external API address to obtain its interface description information. Next, the system parses this description information, automatically identifying the network request protocol (such as HTTP or HTTPS) and parameter format (such as JSON or XML). Subsequently, the identified protocol type and parameter format are mapped to predefined service source categories, matching and labeling the service's source attribute as "external integration service." Simultaneously, combined with the supply chain business knowledge base, the system automatically generates recommended business operation data based on the parsed API functional semantics, such as recommending "reimbursement form invoice verification" as its business operation tag, greatly improving registration efficiency and accuracy.

[0031] Step 102: After the developers confirm, generate complete service registration information and store the service registration information in the database to build a service profile.

[0032] In this embodiment, all attributes extracted and recommended by AI are displayed through a visual interface, allowing developers to perform final review and fine-tuning. This ensures both the efficiency of automation and the reliability of final human control. For example, after developer confirmation, the system integrates all attributes to generate a complete service registration record. This record is a structured data object that includes not only basic attributes such as service number, parameter type, and business operation, but may also include management attributes such as creator, creation time, and service status (e.g., enabled / disabled).

[0033] Specifically, the system persistently stores this complete registration information in a database (such as MySQL). Understandably, the direct purpose of this storage is for subsequent service retrieval and invocation, while a deeper purpose is to build service profiles. A service profile is a dynamic, multi-dimensional service characteristic model. Initially built based on the registration information, it includes static attributes such as the service's functional description and its business domain. In subsequent steps, dynamic runtime metrics such as service call frequency and response time are continuously incorporated into this profile, thus giving each service a comprehensive digital model depicting its own characteristics and behaviors.

[0034] Step 103: In response to the developer's service call request, the machine learning model trained based on historical call data automatically performs data format conversion and protocol adaptation operations for the call parameters.

[0035] In this embodiment, when a developer orchestrates business logic processes on a low-code platform and triggers a "call request" for a registered service, the data format conversion and protocol adaptation of the call parameters are automatically performed.

[0036] It's important to note that service call requests contain parameter data that needs to be passed to the target service. If the target service is an external API, and the calling source is a Java method within the platform, there may be mismatches in parameter formats and communication protocols. To address this issue, the system automatically performs the adaptation process using a machine learning model trained on historical call data. Specifically, the system collects historical service call records as training samples. These records contain the source parameter format, the target parameter format, and the mapping relationship of successful conversions. Subsequently, using the source parameter format as input and the target parameter format as output, machine learning models such as Random Forest are trained.

[0037] For example, when a new service call request is triggered, the system inputs the source parameter format of the current call into the pre-trained model, and the model automatically outputs the predicted target parameter format. The system then performs operations such as serialization (e.g., converting a Java object to a JSON string) or deserialization on the parameter object based on this prediction result, thereby achieving automatic data format conversion and protocol adaptation. For instance, in a supply chain order creation scenario, order objects within the platform can be automatically converted into a specific JSON structure required by the external logistics system, without requiring developers to write any adaptation code, significantly improving call success rate and development efficiency.

[0038] Step 104: Collect service call logs in real time and identify abnormal call events based on a preset anomaly detection model, so as to push the anomaly warning information to the corresponding developers.

[0039] In this embodiment, the system "collects" detailed "logs" generated during each service call in real time. These log data include call timestamps, response times, and return status codes. First, the system continuously collects service call response time data to form a time-series dataset. Then, a Long Short-Term Memory (LSTM) network model is used to train this dataset to learn the normal fluctuation patterns of service calls, thereby constructing a service operation status baseline. This baseline defines the normal range of response times within a specific time window.

[0040] Specifically, during the real-time monitoring phase, the system compares the response time of the current call with a baseline. When the response time consistently exceeds a preset baseline threshold, the call is marked as an abnormal call event. Understandably, to further clarify the cause of the failure, the system extracts multi-dimensional feature data at the time of the abnormal event and inputs it into a pre-trained Gradient Boosting Decision Tree (XGBoost) classification model for inference. This model outputs the fault type corresponding to the abnormal event, such as "network timeout" or "parameter mismatch." Finally, the system immediately pushes anomaly warning information, including the fault type and service identifier, to the corresponding developers, enabling them to respond quickly and significantly shortening fault recovery time, thus ensuring the stable operation of the supply chain business system.

[0041] Step 105: Construct user profiles based on developer information. Based on user profiles and service profiles, use recommendation algorithms to match and recommend suitable service sets to developers, and allow developers to locate target services through semantic search.

[0042] In this embodiment, the system "builds user profiles," with data derived from developers' historical behavior on the platform, such as frequently developed business areas and frequently invoked service types. For example, the intelligent recommendation engine collaboratively analyzes both user and service profiles. Through recommendation algorithms such as collaborative filtering, it calculates the matching degree between the user and all services. For instance, an engineer primarily developing the expense reimbursement module, after completing the "create expense report" service call, will be recommended services such as "invoice verification" based on their profile association, forming a suitable service set.

[0043] Specifically, the system also allows developers to locate target services through semantic search. It receives natural language queries from developers, performs word segmentation and stop word removal to obtain a keyword set; calculates the similarity between this set and the service function description text, such as constructing a vector based on term frequency-inverse document frequency and calculating cosine similarity; finally, it returns the most relevant services based on similarity ranking. It's worth noting that the combination of semantic search and intelligent recommendation significantly lowers the barrier to service retrieval and improves development efficiency.

[0044] In this embodiment, it should first be noted that the system periodically collects feedback data from developers through the user interface at preset time intervals. For example, the preset time interval can be set weekly or monthly, thus ensuring data timeliness while avoiding excessively frequent interruptions. The "user interface" serves as a bridge between the system and developers, and is designed to proactively pop up lightweight feedback forms or provide convenient feedback entry points next to service calls, recommendations, and alert results, such as "thumbs-up / thumbs-down" buttons or short comment boxes.

[0045] The collected feedback data has a clear business focus and mainly includes three categories: first, recommendation accuracy rating data, which is a quantitative evaluation by developers of whether the services recommended by the system are truly useful; second, alert timeliness comment data, which is a textual description by operations and maintenance personnel of whether abnormal alert information is timely and accurate; and third, service usage problem description data, such as specific problem records encountered by developers when trying to use a service generated by AI parsing, including parameter errors or functional mismatches. Understandably, this proactive and structured feedback collection mechanism is key to transforming subjective user experience into objective data that can be used by the model.

[0046] Subsequently, the system automatically integrates the feedback data with the corresponding original service data and call data to form a new training sample set. Specifically, this is a data association and enhancement process. For example, when receiving feedback about "inaccurate invoice verification service recommendations," the system traces the logs to find the context information that triggered the recommendation, including the developer's user profile (such as their project role), the business scenario they were working on at the time, and the original service data such as the service profile of the recommended service. Simultaneously, it also correlates with call data from all users accessing the service over a period of time, such as success rate and response time. By binding the feedback tags to this rich contextual data, the originally singular feedback becomes a high-quality new training sample with a clear optimization direction. These new samples, when combined, more realistically reflect the shortcomings of the current system and changes in actual business needs.

[0047] Finally, the system periodically performs incremental training on the semantic parsing model, service recommendation model, and anomaly detection model using new training sample sets. For example, incremental training is an efficient model update strategy. Instead of retraining the model from scratch, it fine-tunes the existing model parameters using a new training sample set. This approach allows for the absorption of new knowledge while retaining previously learned general patterns, resulting in low computational cost and high efficiency. For instance, the service recommendation model can gradually refine its recommendation strategy by absorbing a large amount of rating data and corresponding context, making it more aligned with the actual preferences of different developers. The semantic parsing model can improve its accuracy in understanding complex method annotations by learning from parsing error cases mentioned in the problem description data. Understandably, after model training is complete, the optimized model must be redeployed to the production environment to complete the entire optimization loop.

[0048] The above are embodiments of the method proposed in this application. Based on the same inventive concept, embodiments of this application also provide a supply chain low-code platform service management device, the structure of which is as follows: Figure 2 As shown.

[0049] Figure 2 This is a schematic diagram of the internal structure of a supply chain low-code platform service management device provided as an embodiment of this application. Figure 2 As shown, the device includes: At least one processor; And, a memory that is communicatively connected to at least one processor; The memory stores instructions that can be executed by at least one processor, and the instructions, when executed by at least one processor, enable at least one processor to: It receives business JAR file uploaded by developers or inputs external API address information, and parses the Java method semantic information in the JAR file to generate a standardized service number, or parses the protocol type and parameter format of the external API address information to extract service attribute data to generate business operation recommendation data; After the developers confirm the information, complete service registration information is generated and stored in the database to build a service profile. In response to service call requests from developers, the machine learning model trained based on historical call data automatically performs data format conversion and protocol adaptation operations for the call parameters; Collect service call logs in real time and identify abnormal call events based on a preset anomaly detection model to push anomaly warning information to the corresponding developers; User profiles are built based on developer information. Based on user profiles and service profiles, a recommendation algorithm is used to match and recommend suitable service sets to developers, and developers are allowed to locate target services through semantic search.

[0050] This application also provides a non-volatile computer storage medium storing computer-executable instructions, which, when executed, can: It receives business JAR file uploaded by developers or inputs external API address information, and parses the Java method semantic information in the JAR file to generate a standardized service number, or parses the protocol type and parameter format of the external API address information to extract service attribute data to generate business operation recommendation data; After the developers confirm the information, complete service registration information is generated and stored in the database to build a service profile. In response to service call requests from developers, the machine learning model trained based on historical call data automatically performs data format conversion and protocol adaptation operations for the call parameters; Collect service call logs in real time and identify abnormal call events based on a preset anomaly detection model to push anomaly warning information to the corresponding developers; User profiles are built based on developer information. Based on user profiles and service profiles, a recommendation algorithm is used to match and recommend suitable service sets to developers, and developers are allowed to locate target services through semantic search.

[0051] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the description of the method embodiments.

[0052] The devices and media provided in this application are one-to-one with the methods. Therefore, the devices and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0053] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0054] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0055] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0056] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0057] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0058] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0059] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0060] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0061] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for managing low-code platform services in a supply chain, characterized in that, The method includes: Receive business JAR package files uploaded by developers or input external API address information, and parse the Java method semantic information in the JAR package file to generate a standardized service number, or parse the protocol type and parameter format of the external API address information to extract service attribute data to generate business operation recommendation data; After the developer confirms the information, complete service registration information is generated and stored in the database to build a service profile. In response to the developer's service call request, the machine learning model trained based on historical call data automatically performs data format conversion and protocol adaptation operations for the call parameters; Collect service call logs in real time and identify abnormal call events based on a preset anomaly detection model to push anomaly warning information to the corresponding developers; User profiles are built based on developer information. Based on the user profiles and service profiles, a recommendation algorithm is used to match and recommend a set of suitable services to the developers, and the developers are allowed to locate target services through semantic search.

2. The supply chain low-code platform service management method according to claim 1, characterized in that, Parsing the Java method semantic information in the aforementioned Jar package file to generate standardized service numbers specifically includes: Load the business JAR file, and traverse and parse the Java class and method definitions in the business JAR file; Using a pre-trained language model based on the Transformer architecture, the parsed Java method source code is encoded to generate high-dimensional semantic vectors; Analyze the high-dimensional semantic vector to automatically identify and extract the method name, the parameter types that constitute the method input, and the return value type that constitutes the method output, which represent the function of the method. Based on predefined number generation rules, the extracted attribute data is concatenated to generate a standardized service number string.

3. The supply chain low-code platform service management method according to claim 1, characterized in that, Parse the protocol type and parameter format of the external API address information to extract service attribute data, and generate business operation recommendation data, specifically including: Send a probe request to the external API address to obtain the interface description information of the external API address; The interface description information is parsed to automatically identify the protocol type corresponding to the network request protocol used by the external API address; the protocol type includes HTTP or HTTPS. The request body and response body structure definitions in the interface description information are parsed to automatically identify the parameter format of the external API address, wherein the parameter format includes JSON or XML; The identified protocol types and parameter formats are mapped to predefined service source categories, and the service source attributes are matched and labeled as external integration services.

4. The supply chain low-code platform service management method according to claim 1, characterized in that, A machine learning model trained on historical call data automatically performs data format conversion and protocol adaptation operations for call parameters, specifically including: Collect historical service call records and use these records as training samples to train a random forest machine learning model, with the source parameter format as input and the target parameter format as output; the historical service call records contain the source parameter format, the target parameter format, and the mapping relationship of successful conversion; When a new service call request is triggered, the source parameter format of the current call is input into the trained random forest model, and the predicted target parameter format is output. Based on the predicted target parameter format, serialization or deserialization operations are performed on the parameter object to achieve data format conversion and protocol adaptation.

5. A supply chain low-code platform service management method according to claim 1, characterized in that, Real-time collection of service call logs, and identification of abnormal call events based on a pre-defined anomaly detection model, specifically including: Continuously collect service call response time data to form a time series dataset; The time series dataset is trained using a Long Short-Term Memory (LSTM) network model to learn normal fluctuation patterns in service calls in order to construct a service operation status baseline; the service operation status baseline includes a predefined normal range of response time within a specific time window; The call response time of the real-time monitoring service is compared with the service operation status baseline. When the call response time continuously exceeds the preset baseline threshold range, the current call event is marked as an abnormal call event. Extract the multidimensional feature data when the abnormal call event occurs, and input the multidimensional feature data into a pre-trained gradient boosting decision tree classification model for inference, and output the fault type corresponding to the abnormal call event.

6. The supply chain low-code platform service management method according to claim 1, characterized in that, The developers are allowed to locate target services through semantic search, specifically including: Receive natural language query statements input by developers, and perform word segmentation and stop word removal on the natural language query statements to obtain a keyword set; Calculate the term frequency inverse document frequency (TNV) values ​​of the keyword set and all service function description texts in the service profile, and construct the text vectors of the corresponding keywords based on the TNV values; Calculate the cosine similarity between the text vector corresponding to each keyword in the keyword set and the service description vector corresponding to the service function description text, and sort the services in the service profile according to the cosine similarity, so as to return and locate the service with the highest similarity in the sorted list as the target service.

7. A supply chain low-code platform service management method according to claim 1, characterized in that, Before receiving the business JAR file uploaded by the developer or the external API address information entered, the method further includes: The service management system is deployed on the application server of the supply chain low-code platform, and a database instance is integrated to persist storage service metadata and call historical records and model training results. Import scenario tag data and historical service data from the supply chain business knowledge base to form a training dataset; the scenario tag data includes semantic descriptions of expense reimbursement management, logistics traceability and order fulfillment, and the historical service data includes service numbers, parameter types, call frequencies and error logs of registered services; Based on the training dataset, natural language processing techniques are used to train a semantic parsing model, a service recommendation model, and an anomaly detection model, and the parameters of the semantic parsing model, the service recommendation model, and the anomaly detection model are initialized.

8. A supply chain low-code platform service management method according to claim 1, characterized in that, The method further includes: Feedback data from developers is collected periodically through the user interface at preset time intervals, and the feedback data is integrated with the corresponding original service data and call data to form a new training sample set; the feedback data includes recommendation accuracy score data, warning timeliness comment data, and service usage problem description data; The semantic parsing model, service recommendation model, and anomaly detection model are incrementally trained periodically using new training sample sets, and the optimized models are redeployed to the production environment.

9. A supply chain low-code platform service management device, characterized in that, The device includes: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform a supply chain low-code platform service management method as described in any one of claims 1-8.

10. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, When the computer-executable instructions are executed, a supply chain low-code platform service management method as described in any one of claims 1-8 is implemented.

Images