Multi-language compatible algorithm scheduling platform and method

Abstract

The invention provides a multi-language compatible algorithm scheduling platform and method, and the platform comprises a data collection module which is used for receiving basic information and algorithm files of different language algorithms based on an API interface, and converting the algorithm files into mirror image files according to the basic information based on a preset method; the algorithm deployment module is used for carrying out environment configuration on the algorithm container based on the basic information, carrying out isolated deployment on the mirror image file in the algorithm container based on an environment configuration result, and carrying out multi-dimensional management on different language algorithms subjected to isolated deployment; and the algorithm scheduling module is used for receiving a service demand based on the multi-dimensional management result, locking a target language algorithm based on the service demand, and performing scheduling operation on the target language algorithm based on an API interface. According to the method, different language algorithms can stably and reliably run in a deployment environment, the reliability of cross-language deployment is improved, and the reliability, stability and accuracy of multi-language algorithm scheduling are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of algorithm scheduling, and in particular to a multi-language compatible algorithm scheduling platform and method. Background Art

[0002] At present, with the rapid development of data science and artificial intelligence technology, the application of power prediction algorithm models is becoming more and more extensive. These models are often implemented in a variety of programming languages. For example, Python has become a popular choice for data preprocessing and model training due to its rich scientific computing library. R has a deep accumulation in the field of statistical analysis, and Java is often used in the development of backend services due to its high performance and cross-platform capabilities.

[0003] However, the coexistence of multiple languages ​​also brings many challenges. First, the interoperability between different programming languages ​​is poor, which leads to many difficulties in cross-language deployment of models. If the target environment does not install or does not support the programming language used by the model, the model will not be able to run in this environment. Secondly, even if the target environment supports the required programming language, the model may not work properly due to version incompatibility issues, because different versions of programming languages ​​may have differences in syntax, library functions or APIs, and these differences may directly affect the execution results of the model.

[0004] Therefore, in order to overcome the above-mentioned defects, the present invention provides a multi-language compatible algorithm scheduling platform and method. Summary of the invention

[0005] The present invention provides a multi-language compatible algorithm scheduling platform and method, which are used to receive basic information and algorithm files of different language algorithms, and convert the algorithm files into mirror files according to a preset method, so as to facilitate the deployment operation of different language algorithms. Secondly, the algorithm container is configured according to the basic information of the different language algorithms, so as to ensure that the different language algorithms can run stably and reliably in the deployment environment. At the same time, the different language algorithms are deployed in isolation in the algorithm container, so as to avoid mutual interference and resource competition, and improve the reliability of cross-language deployment. After successful deployment, the different language algorithms are managed in multiple dimensions, so as to facilitate timely update and deletion of different language algorithms, and improve the management effect of different language algorithms. Finally, the business needs are responded to according to the multi-dimensional management results, and the target language algorithm is accurately matched, and the target language algorithm is accurately and effectively scheduled, so as to improve the reliability, stability and accuracy of multi-language algorithm scheduling.

[0006] The present invention provides a multi-language compatible algorithm scheduling platform, comprising:

[0007] The data collection module is used to receive basic information and algorithm files of different language algorithms based on the API interface, and convert the algorithm files into image files according to the basic information based on a preset method;

[0008] The algorithm deployment module is used to configure the environment of the algorithm container based on basic information, and to deploy the image file in the algorithm container in isolation based on the environment configuration result, and to perform multi-dimensional management of the isolated and deployed algorithms of different languages;

[0009] The algorithm scheduling module is used to receive business requirements based on multi-dimensional management results, lock the target language algorithm based on the business requirements, and schedule the target language algorithm based on the API interface.

[0010] Preferably, a multi-language compatible algorithm scheduling platform, data collection module, comprises:

[0011] The interface configuration unit is used to obtain the parameter configuration document of the API interface, parse the parameter configuration document, determine the basic configuration parameters of the API interface, and build an adaptation layer for different language algorithms based on the basic configuration parameters;

[0012] An information receiving unit, used to obtain basic information and algorithm files of different language algorithms entered by the user based on the user interface, and send the basic information and algorithm files of different language algorithms to the API interface;

[0013] The data preprocessing unit is used to verify the basic information and algorithm files of different language algorithms based on the API interface, and format the algorithm files of different language algorithms based on the adaptation layer after the verification, and cache the formatted different language algorithms.

[0014] Preferably, a multi-language compatible algorithm scheduling platform, a data preprocessing unit, comprises:

[0015] The permission verification subunit is used to:

[0016] Obtain data sources of different language algorithms based on the API interface, and determine the identity tags of device terminals corresponding to different language algorithms based on the data sources;

[0017] The identity tag is matched with the preset permission library, and the upload permission of the device terminal corresponding to the different language algorithms is determined based on the data matching result. If the upload permission exists, the corresponding language algorithm is received based on the API interface, otherwise, the corresponding language algorithm is returned;

[0018] The data verification sub-unit is used to verify the illegal characters and execution logic of different language algorithms based on the protocol texts of different language algorithms after the upload permission verification is passed, and input the different language algorithms into the adaptation layer for processing after the illegal characters and execution logic are verified.

[0019] Preferably, a multi-language compatible algorithm scheduling platform, data collection module, comprises:

[0020] Image preparation unit, used to:

[0021] Obtain basic information about different language algorithms, and match base images from the preset image library based on the basic information;

[0022] At the same time, the operating requirements of different language algorithms are determined based on the basic information, and the corresponding necessary environment variables are determined based on the operating requirements;

[0023] Image file generation unit, used for:

[0024] Add algorithm files of different language algorithms to the base image, and configure the base image with the added algorithm files based on the necessary environment variables;

[0025] Based on the configuration result, a preset image file generation mechanism is started, and based on the preset image file generation mechanism, the configuration result of the basic image with the algorithm file added is processed in a process-cooperative manner to obtain a final image file.

[0026] Preferably, a multi-language compatible algorithm scheduling platform, an algorithm deployment module, comprises:

[0027] Environment hive for:

[0028] Determine the application scenarios of different language algorithms based on their basic information, and determine the runtime libraries and dependent software required by different language algorithms based on the application scenarios;

[0029] Bind the runtime library and dependent software to the corresponding algorithm container, and complete the environment configuration of the algorithm container based on the binding results;

[0030] Isolate deployment units for:

[0031] Deploy different language algorithms in corresponding algorithm containers, and determine the current available resources and the target resources required for each algorithm container to run based on the deployment results;

[0032] Adaptively allocate resource loads to each algorithm container based on available resources and target resources required for each algorithm container to run. At the same time, determine the importance parameters of different language algorithms based on basic information, and set container permissions for corresponding algorithm containers based on the adaptive allocation results and importance parameters to complete isolation deployment.

[0033] Multi-dimensional management unit for:

[0034] Based on the isolated deployment results, the update status of the corresponding language algorithm in each algorithm container is monitored in real time, and when there is an algorithm update, the algorithm update request and algorithm update file are received;

[0035] Parse the algorithm update file based on the algorithm update request, determine the update parameters and the update nodes of the old version of the algorithm, and replace the parameters of the update nodes of the old version of the algorithm based on the update parameters;

[0036] Generate an updated version of the algorithm based on the parameter replacement result, and authorize the updated version of the algorithm. At the same time, determine the data difference characteristics between the updated version of the algorithm and the old version of the algorithm, and determine the adaptive adjustment strategy for the algorithm container based on the data difference characteristics;

[0037] The algorithm container is dynamically configured based on the adaptive adjustment strategy, and after the adjustment is completed, the old version of the algorithm is recorded and archived in sequence based on the update time node.

[0038] Preferably, a multi-language compatible algorithm scheduling platform and a multi-dimensional management unit include:

[0039] A result acquisition subunit is used to obtain the isolation deployment results obtained, and determine the index tags corresponding to different algorithm containers based on the isolation deployment results;

[0040] Management subunit, used to:

[0041] Summarize the algorithm containers of different language algorithms to obtain algorithm container clusters, and build a search engine for the algorithm container clusters. At the same time, add index tags of different algorithm containers to the background script file of the search engine.

[0042] Respond to the search request submitted by the user based on the added results, and locate the language algorithm to be managed based on the response results;

[0043] An active management request submitted by a user to the managed language algorithm is obtained, and based on the active management request, a management mechanism is triggered to perform corresponding management operations on the managed language algorithm, wherein the active management request includes active update and deletion.

[0044] Preferably, a multi-language compatible algorithm scheduling platform, an algorithm scheduling module, comprises:

[0045] The demand receiving unit is used to receive the service demand sent by the user terminal and convert the service demand into text to obtain the demand data;

[0046] A key data determination unit is used to perform semantic analysis on the demand data based on a neural network, obtain semantic features corresponding to the demand data, and determine key tasks and constraints of the business requirements based on the semantic features;

[0047] An algorithm determination unit is used to extract the business features of different language algorithms, determine the subordination of key tasks and constraints, and lock the target language algorithm based on the subordination;

[0048] The algorithm scheduling unit is used to schedule the target algorithm based on the API interface, and at the same time, synchronously schedule the resources required by the target algorithm.

[0049] Preferably, a multi-language compatible algorithm scheduling platform, an algorithm scheduling unit, comprises:

[0050] The condition setting subunit is used to set the preset trigger condition in the background and determine the associated factors included in the preset trigger condition;

[0051] The automatic scheduling subunit is used to monitor the status of the associated factors in real time and trigger the scheduling operation of the required language algorithm when the status of the associated factors meets the preset trigger conditions;

[0052] The recording subunit is used to collect the basic information and scheduling timestamp of the language algorithm corresponding to each scheduling operation based on the trigger result, and record the basic information and scheduling timestamp of the language algorithm corresponding to each scheduling operation, generate an algorithm scheduling log, and record the algorithm scheduling log.

[0053] The present invention provides a multi-language compatible algorithm scheduling method, comprising:

[0054] Step 1: Receive basic information and algorithm files of different language algorithms based on the API interface, and convert the algorithm files into image files according to the basic information based on the preset method;

[0055] Step 2: Configure the environment of the algorithm container based on the basic information, and deploy the image file in the algorithm container in isolation based on the environment configuration results, and manage the isolated and deployed different language algorithms in multiple dimensions;

[0056] Step 3: Receive business requirements based on the multi-dimensional management results, lock the target language algorithm based on the business requirements, and schedule the target language algorithm based on the API interface.

[0057] Preferably, a multi-language compatible algorithm scheduling method, step 1: receiving basic information and algorithm files of algorithms in different languages ​​based on an API interface, including:

[0058] Obtain the parameter configuration document of the API interface, parse the parameter configuration document, determine the basic configuration parameters of the API interface, and build an adaptation layer for different language algorithms based on the basic configuration parameters;

[0059] Obtain the basic information and algorithm files of different language algorithms entered by the user based on the user interface, and send the basic information and algorithm files of different language algorithms to the API interface;

[0060] The basic information and algorithm files of different language algorithms are verified based on the API interface, and after the verification, the algorithm files of different language algorithms are formatted based on the adaptation layer, and the formatted different language algorithms are cached.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. By receiving the basic information and algorithm files of different language algorithms, and converting the algorithm files into image files according to the preset method, it is convenient to deploy different language algorithms. Secondly, the algorithm container is configured according to the basic information of different language algorithms, so as to ensure that different language algorithms can run stably and reliably in the deployment environment. At the same time, different language algorithms are deployed in isolation in the algorithm container to avoid mutual interference and resource competition, thereby improving the reliability of cross-language deployment. After successful deployment, different language algorithms are managed in multiple dimensions, which facilitates timely updating and deletion of different language algorithms, thereby improving the management effect of different language algorithms. Finally, respond to business needs according to the multi-dimensional management results, accurately match the target language algorithm, and accurately and effectively schedule the target language algorithm, thereby improving the reliability, stability and accuracy of multi-language algorithm scheduling.

[0063] 2. By parsing the parameter configuration document of the API interface, the basic configuration parameters of the API interface can be accurately and effectively determined, so that the adaptation layer of different language algorithms can be accurately and effectively constructed according to the basic configuration parameters, which provides convenience for the effective reception of different language algorithms. Secondly, the basic information and algorithm files of different language algorithms are received through the user interface, so as to facilitate the effective acquisition of the basic information and specific content of different language algorithms. Finally, the basic information and algorithm files of different language algorithms received are verified, and after the verification, the different language algorithms are formatted and cached, which provides convenience for the deployment and scheduling of different language algorithms.

[0064] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in this application document.

[0065] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0067] Figure 1 It is a structural diagram of a multi-language compatible algorithm scheduling platform in an embodiment of the present invention;

[0068] Figure 2 It is a structural diagram of a data collection module in a multi-language compatible algorithm scheduling platform in an embodiment of the present invention;

[0069] Figure 3 The present invention is a flowchart of a multi-language compatible algorithm scheduling method in an embodiment of the present invention. DETAILED DESCRIPTION

[0070] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0071] Embodiment 1:

[0072] This embodiment provides a multi-language compatible algorithm scheduling platform, such as Figure 1 As shown, including:

[0073] The data collection module is used to receive basic information and algorithm files of different language algorithms based on the API interface, and convert the algorithm files into image files according to the basic information based on a preset method;

[0074] The algorithm deployment module is used to configure the environment of the algorithm container based on basic information, and to deploy the image file in the algorithm container in isolation based on the environment configuration result, and to perform multi-dimensional management of the isolated and deployed algorithms of different languages;

[0075] The algorithm scheduling module is used to receive business requirements based on multi-dimensional management results, lock the target language algorithm based on the business requirements, and schedule the target language algorithm based on the API interface.

[0076] In this embodiment, the API interface (application programming interface) is a set of rules and specifications that define the way software components interact with each other, allowing different software systems or applications to communicate and exchange data. Through the API interface, one software can provide specific functions or data access rights to other software, and other software can call these functions or obtain data in the manner defined by the interface.

[0077] In this embodiment, different language algorithms refer to algorithms constructed in Python, Java, JavaScript, etc.

[0078] In this embodiment, the basic information refers to the name, version number, detailed description, author information, technical field, external library dependency, and precise input and output specifications of different language algorithms.

[0079] In this embodiment, the algorithm file refers to a file containing specific contents corresponding to different language algorithms.

[0080] In this embodiment, the preset method is known in advance, for example, it can be a Docker image.

[0081] In this embodiment, the image file refers to a file that can be directly deployed in the algorithm container, including the content and structure information of the entire language algorithm.

[0082] In this embodiment, environment configuration refers to the configuration of software and databases required for the algorithm container to run, wherein the algorithm container is a space for carrying language algorithms.

[0083] In this embodiment, isolated deployment refers to deploying different language algorithms in different algorithm containers, thereby avoiding mutual interference and resource competition.

[0084] In this embodiment, multi-dimensional management refers to management operations such as updating, deleting, and retrieving different language algorithms.

[0085] In this embodiment, the business demand refers to the data content that the user currently needs to process.

[0086] In this embodiment, the target language algorithm refers to an algorithm among the deployed different language algorithms that can solve the current business needs.

[0087] The working principle and beneficial effects of the above technical solution are: by receiving the basic information and algorithm files of different language algorithms, and converting the algorithm files into image files according to a preset method, it is convenient to deploy different language algorithms. Secondly, the algorithm container is configured according to the basic information of different language algorithms, thereby ensuring that different language algorithms can run stably and reliably in the deployment environment. At the same time, different language algorithms are deployed in isolation in the algorithm container to avoid mutual interference and resource competition, thereby improving the reliability of cross-language deployment. After successful deployment, different language algorithms are managed in multiple dimensions, which facilitates timely updating and deletion of different language algorithms, thereby improving the management effect of different language algorithms. Finally, business needs are responded to according to the multi-dimensional management results, accurate matching of target language algorithms is achieved, and accurate and effective scheduling of target language algorithms is performed, thereby improving the reliability, stability and accuracy of multi-language algorithm scheduling.

[0088] Embodiment 2:

[0089] Based on Example 1, this example provides a multi-language compatible algorithm scheduling platform, such as Figure 2 As shown, the data collection module includes:

[0090] The interface configuration unit is used to obtain the parameter configuration document of the API interface, parse the parameter configuration document, determine the basic configuration parameters of the API interface, and build an adaptation layer for different language algorithms based on the basic configuration parameters;

[0091] An information receiving unit, used to obtain basic information and algorithm files of different language algorithms entered by the user based on the user interface, and send the basic information and algorithm files of different language algorithms to the API interface;

[0092] The data preprocessing unit is used to verify the basic information and algorithm files of different language algorithms based on the API interface, and format the algorithm files of different language algorithms based on the adaptation layer after the verification, and cache the formatted different language algorithms.

[0093] In this embodiment, the parameter configuration document refers to a file that records the data format and working parameter definitions of the API interface.

[0094] In this embodiment, the basic configuration parameters refer to parameters such as specific data input and output formats corresponding to the API interface.

[0095] In this embodiment, the adaptation layer is a mechanism for unifying the formats of different language algorithms.

[0096] In this embodiment, the user interface is built in advance, and the user can enter the basic information and algorithm files of different language algorithms through the user interface.

[0097] In this embodiment, verifying the basic information and algorithm files of different language algorithms based on the API interface refers to verifying the authority and data standardization of the basic information and algorithm files of different language algorithms.

[0098] The working principle and beneficial effects of the above technical solution are: by parsing the parameter configuration document of the API interface, the basic configuration parameters of the API interface can be accurately and effectively determined, so as to realize the accurate and effective construction of the adaptation layer of different language algorithms according to the basic configuration parameters, which provides convenience for the effective reception of different language algorithms. Secondly, the basic information and algorithm files of different language algorithms are received through the user interface, so as to facilitate the effective acquisition of the basic information and specific content of different language algorithms. Finally, the basic information and algorithm files of the received different language algorithms are verified, and the different language algorithms are formatted and cached after the verification, which provides convenience for the deployment and scheduling of different language algorithms.

[0099] Embodiment 3:

[0100] Based on Example 2, this embodiment provides a multi-language compatible algorithm scheduling platform, a data preprocessing unit, including:

[0101] The permission verification subunit is used to:

[0102] Obtain data sources of different language algorithms based on the API interface, and determine the identity tags of device terminals corresponding to different language algorithms based on the data sources;

[0103] The identity tag is matched with the preset permission library, and the upload permission of the device terminal corresponding to the different language algorithms is determined based on the data matching result. If the upload permission exists, the corresponding language algorithm is received based on the API interface, otherwise, the corresponding language algorithm is returned;

[0104] The data verification subunit is used to verify the illegal characters and execution logic of different language algorithms based on the protocol texts of different language algorithms after the upload permission verification is passed, and input the different language algorithms into the adaptation layer for processing after the illegal characters and execution logic are verified.

[0105] In this embodiment, the data source refers to the terminal corresponding to different language algorithms.

[0106] In this embodiment, the identity tag refers to a marking symbol that can distinguish different device terminals.

[0107] In this embodiment, the preset permission library is known in advance and is used to record the upload permissions of different device terminals for the language algorithm.

[0108] In this embodiment, the protocol text refers to the conditions that different language algorithms need to meet when working and the corresponding execution logic when running.

[0109] In this embodiment, the illegal characters refer to programming characters that cannot appear in the corresponding systems of different language algorithms.

[0110] In this embodiment, the execution logic refers to the execution order between various links of different language algorithms during runtime.

[0111] The working principle and beneficial effects of the above technical solution are: by effectively acquiring the identity tags of the corresponding device terminals according to the data sources of different language algorithms, it is convenient to effectively verify the upload permissions of different device terminals to the language algorithms according to the identity tags; secondly, after the upload permission verification is passed, the illegal characters and execution logic of different language algorithms are verified to ensure the standardization and reliability of different language algorithms; finally, after the verification is passed, they are input into the adaptation layer for processing to ensure the reliability of different language algorithms.

[0112] Embodiment 4:

[0113] Based on Example 1, this embodiment provides a multi-language compatible algorithm scheduling platform, a data collection module, including:

[0114] Image preparation unit, used to:

[0115] Obtain basic information about different language algorithms, and match base images from the preset image library based on the basic information;

[0116] At the same time, the operating requirements of different language algorithms are determined based on the basic information, and the corresponding necessary environment variables are determined based on the operating requirements;

[0117] Image file generation unit, used for:

[0118] Add algorithm files of different language algorithms to the base image, and configure the base image with the added algorithm files based on the necessary environment variables;

[0119] Based on the configuration result, a preset image file generation mechanism is started, and based on the preset image file generation mechanism, the configuration result of the basic image with the algorithm file added is processed in a process-cooperative manner to obtain a final image file.

[0120] In this embodiment, the output image library is known in advance, and contains a plurality of different types of basic images.

[0121] In this embodiment, the base image refers to a framework that can convert different programs into corresponding image files.

[0122] In this embodiment, the operation requirements refer to the environment requirements and configuration requirements that different language algorithms need to meet during operation.

[0123] In this embodiment, necessary environment variables refer to conditions and numerical requirements that must be met by the operating environment when different language algorithms are running.

[0124] In this embodiment, the preset image file generation mechanism is a strategy trained in advance for generating image files.

[0125] The working principle and beneficial effects of the above technical solution are: by matching the corresponding basic image from the preset image library according to the basic information of different language algorithms, at the same time, locking the necessary environment variables according to the running requirements of different language algorithms, secondly, adding the algorithm files of different language algorithms to the basic image, and configuring the basic image with the added algorithm files according to the determined necessary environment variables, and finally, through the preset image file generation mechanism, the configuration results of the basic image with the added algorithm files are processed in a process-coordinated manner, so as to realize the accurate and effective generation of image files, and provide convenience for the scheduling and management of multi-language compatible algorithms.

[0126] Embodiment 5:

[0127] Based on Example 1, this embodiment provides a multi-language compatible algorithm scheduling platform, an algorithm deployment module, including:

[0128] Environment hive for:

[0129] Determine the application scenarios of different language algorithms based on their basic information, and determine the runtime libraries and dependent software required by different language algorithms based on the application scenarios;

[0130] Bind the runtime library and dependent software to the corresponding algorithm container, and complete the environment configuration of the algorithm container based on the binding results;

[0131] Isolate deployment units for:

[0132] Deploy different language algorithms in corresponding algorithm containers, and determine the current available resources and the target resources required for each algorithm container to run based on the deployment results;

[0133] Adaptively allocate resource loads to each algorithm container based on available resources and target resources required for each algorithm container to run. At the same time, determine the importance parameters of different language algorithms based on basic information, and set container permissions for corresponding algorithm containers based on the adaptive allocation results and importance parameters to complete isolation deployment.

[0134] Multi-dimensional management unit for:

[0135] Based on the isolated deployment results, the update status of the corresponding language algorithm in each algorithm container is monitored in real time, and when there is an algorithm update, the algorithm update request and algorithm update file are received;

[0136] Parse the algorithm update file based on the algorithm update request, determine the update parameters and the update nodes of the old version of the algorithm, and replace the parameters of the update nodes of the old version of the algorithm based on the update parameters;

[0137] Generate an updated version of the algorithm based on the parameter replacement result, and authorize the updated version of the algorithm. At the same time, determine the data difference characteristics between the updated version of the algorithm and the old version of the algorithm, and determine the adaptive adjustment strategy for the algorithm container based on the data difference characteristics;

[0138] The algorithm container is dynamically configured based on the adaptive adjustment strategy, and after the adjustment is completed, the old version of the algorithm is recorded and archived in sequence based on the update time node.

[0139] In this embodiment, the application scenario refers to problems that can be solved by different language algorithms or businesses that can be applied.

[0140] In this embodiment, the runtime library refers to a database required by different language algorithms during runtime.

[0141] In this embodiment, dependent software refers to programs that different language algorithms need to rely on during operation.

[0142] In this embodiment, the available resource amount refers to the size of resources that can be effectively used by the algorithm container at present when it is running.

[0143] In this embodiment, the target resource amount refers to the amount of resources required by each algorithm container during normal operation.

[0144] In this embodiment, adaptive allocation refers to the reasonable allocation of currently available resources for each algorithm container based on the currently available resources and the target resources required for the operation of each algorithm container. When allocating, the available resources need to be split and allocated according to the ratio of the target resources required for the operation of each algorithm container.

[0145] In this embodiment, the importance parameter refers to the importance of different language algorithms.

[0146] In this embodiment, the container authority is a parameter used to encrypt the authority of different algorithm containers, in order to ensure that different algorithm containers do not affect each other and operate stably.

[0147] In this embodiment, the algorithm update file refers to a file that records a specific language algorithm when the language algorithm is updated, that is, a file corresponding to the new language algorithm.

[0148] In this embodiment, the update parameter refers to the specific data of the update of the parameters recorded in the algorithm update file relative to the previous language algorithm.

[0149] In this embodiment, the update node of the old version algorithm refers to a specific location where an update operation needs to be performed on the previous language algorithm.

[0150] In this embodiment, the updated version algorithm refers to the final result obtained after the previous language algorithm is updated through the algorithm update file.

[0151] In this embodiment, role authorization refers to authorizing the updated version algorithm obtained through the update, that is, ensuring that the updated version algorithm can replace the old version algorithm to perform corresponding computing operations.

[0152] In this embodiment, the data difference feature refers to the difference between the updated version algorithm and the old version algorithm.

[0153] In this embodiment, the adaptive adjustment strategy refers to an adjustment strategy for the algorithm container determined according to the data difference characteristics, which corresponds to the data difference characteristics and only some parameters of the algorithm container are synchronously adjusted.

[0154] The working principle and beneficial effects of the above technical solution are: by analyzing the basic information of different language algorithms, the runtime libraries and dependent software required by different language algorithms can be effectively determined, so as to realize reliable environment configuration of algorithm containers according to the runtime libraries and dependent software. Secondly, different language algorithms are deployed in corresponding algorithm containers, and the resources required for each algorithm container at runtime are adaptively allocated according to the deployment results, thereby ensuring the reliability of the algorithm container at runtime. At the same time, different language algorithms are deployed in isolation, thereby avoiding mutual interference between different language algorithms. Finally, the update status of different language algorithms is monitored to ensure that when there is a language algorithm update, the corresponding language algorithm and the corresponding algorithm container are accurately and effectively updated, and the old version of the algorithm is recorded and archived in turn, thereby improving the management effect of multi-language algorithms.

[0155] Embodiment 6:

[0156] Based on Example 5, this embodiment provides a multi-language compatible algorithm scheduling platform and a multi-dimensional management unit, including:

[0157] A result acquisition subunit is used to obtain the isolation deployment results obtained, and determine the index tags corresponding to different algorithm containers based on the isolation deployment results;

[0158] Management subunit, used to:

[0159] Summarize the algorithm containers of different language algorithms to obtain algorithm container clusters, and build a search engine for the algorithm container clusters. At the same time, add index tags of different algorithm containers to the background script file of the search engine.

[0160] Respond to the search request submitted by the user based on the added results, and locate the language algorithm to be managed based on the response results;

[0161] An active management request submitted by a user to the managed language algorithm is obtained, and based on the active management request, a management mechanism is triggered to perform corresponding management operations on the managed language algorithm, wherein the active management request includes active update and deletion.

[0162] In this embodiment, the index tag refers to a specific parameter that can locate different language algorithms.

[0163] In this embodiment, the algorithm container cluster refers to a group obtained by aggregating algorithm containers of different language algorithms.

[0164] In this embodiment, the language algorithm to be managed refers to a language algorithm that the user needs to access or call.

[0165] In this embodiment, the active management request refers to a request actively submitted by a user, including operations such as algorithm update and deletion.

[0166] In this embodiment, the active management request triggering management mechanism is trained in advance.

[0167] The working principle and beneficial effects of the above technical solution are: by determining the index tags corresponding to different algorithm containers after isolated deployment, and configuring the constructed search engine according to the determined index tags, it is convenient to respond to retrieval requests according to the configuration results, and secondly, respond to retrieval requests submitted by users, and locate the language algorithm to be managed based on the response results, and at the same time analyze the active management requests submitted by users to the language algorithm to be managed, so as to trigger the management mechanism according to the active management request to perform corresponding management operations on the language algorithm to be managed, thereby ensuring the management effect of different language algorithms.

[0168] Embodiment 7:

[0169] Based on Example 1, this embodiment provides a multi-language compatible algorithm scheduling platform, an algorithm scheduling module, including:

[0170] The demand receiving unit is used to receive the service demand sent by the user terminal and convert the service demand into text to obtain the demand data;

[0171] A key data determination unit is used to perform semantic analysis on the demand data based on a neural network, obtain semantic features corresponding to the demand data, and determine key tasks and constraints of the business requirements based on the semantic features;

[0172] An algorithm determination unit is used to extract the business features of different language algorithms, determine the subordination of key tasks and constraints, and lock the target language algorithm based on the subordination;

[0173] The algorithm scheduling unit is used to schedule the target algorithm based on the API interface, and at the same time, synchronously schedule the resources required by the target algorithm.

[0174] In this embodiment, the demand data refers to the result of converting the service demand sent by the user terminal into specific text or data.

[0175] In this embodiment, the semantic feature refers to the specific meaning corresponding to the demand data.

[0176] In this embodiment, the key tasks and constraints refer to specific business types and condition requirements corresponding to the business needs.

[0177] In this embodiment, the service characteristics refer to the types of services that can be handled by different language algorithms.

[0178] In this embodiment, the degree of subordination refers to the adaptability of the key tasks and constraints to different language algorithms obtained by matching the key tasks and constraints with the business characteristics of different language algorithms.

[0179] In this embodiment, the target language algorithm refers to an algorithm that matches the business requirements.

[0180] The working principle and beneficial effects of the above technical solution are: by parsing the business requirements sent by the user terminal, the key tasks and constraints of the business requirements can be accurately and effectively determined. At the same time, the business characteristics of different language algorithms are extracted, and the key tasks and constraints are matched with the business characteristics of different language algorithms, so as to accurately and effectively determine the target language algorithm. Finally, the target algorithm is scheduled through the API interface, ensuring the reliability and convenience of scheduling different language algorithms.

[0181] Embodiment 8:

[0182] Based on Example 7, this embodiment provides a multi-language compatible algorithm scheduling platform, an algorithm scheduling unit, including:

[0183] The condition setting subunit is used to set the preset trigger condition in the background and determine the associated factors included in the preset trigger condition;

[0184] The automatic scheduling subunit is used to monitor the status of the associated factors in real time and trigger the scheduling operation of the required language algorithm when the status of the associated factors meets the preset trigger conditions;

[0185] The recording subunit is used to collect the basic information and scheduling timestamp of the language algorithm corresponding to each scheduling operation based on the trigger result, and record the basic information and scheduling timestamp of the language algorithm corresponding to each scheduling operation, generate an algorithm scheduling log, and record the algorithm scheduling log.

[0186] In this embodiment, the preset trigger condition is known in advance and is used to represent the corresponding condition or event when the multi-language algorithm is automatically scheduled.

[0187] In this embodiment, the associated factor refers to a condition or parameter related to a preset trigger condition, for example, a period requirement when scheduling a certain language algorithm.

[0188] In this embodiment, the scheduling timestamp refers to the specific time information corresponding to the scheduling operation of different language algorithms.

[0189] The working principle and beneficial effects of the above technical solution are: by setting preset trigger conditions in the background and determining the associated factors included in the preset trigger conditions, real-time monitoring of the status of the associated factors is achieved, so that when the speech trigger conditions are met, the required language algorithm is automatically scheduled, thereby improving the intelligence of the algorithm scheduling. At the same time, the scheduling timestamps and basic information of different language algorithms are recorded to achieve effective generation of algorithm scheduling logs, which is convenient for effective understanding of the scheduling status of different language algorithms.

[0190] Embodiment 9:

[0191] This embodiment provides a multi-language compatible algorithm scheduling method, such as Figure 3 As shown, including:

[0192] Step 1: Receive basic information and algorithm files of different language algorithms based on the API interface, and convert the algorithm files into image files according to the basic information based on the preset method;

[0193] Step 2: Configure the environment of the algorithm container based on the basic information, and deploy the image file in the algorithm container in isolation based on the environment configuration results, and manage the isolated and deployed different language algorithms in multiple dimensions;

[0194] Step 3: Receive business requirements based on the multi-dimensional management results, lock the target language algorithm based on the business requirements, and schedule the target language algorithm based on the API interface.

[0195] The working principle and beneficial effects of the above technical solution are: by receiving the basic information and algorithm files of different language algorithms, and converting the algorithm files into image files according to a preset method, it is convenient to deploy different language algorithms. Secondly, the algorithm container is configured according to the basic information of different language algorithms, thereby ensuring that different language algorithms can run stably and reliably in the deployment environment. At the same time, different language algorithms are deployed in isolation in the algorithm container to avoid mutual interference and resource competition, thereby improving the reliability of cross-language deployment. After successful deployment, different language algorithms are managed in multiple dimensions, which facilitates timely updating and deletion of different language algorithms, thereby improving the management effect of different language algorithms. Finally, business needs are responded to according to the multi-dimensional management results, accurate matching of target language algorithms is achieved, and accurate and effective scheduling of target language algorithms is performed, thereby improving the reliability, stability and accuracy of multi-language algorithm scheduling.

[0196] Embodiment 10:

[0197] Based on Example 9, this example provides a multi-language compatible algorithm scheduling method, step 1: receiving basic information and algorithm files of algorithms in different languages ​​based on an API interface, including:

[0198] Obtain the parameter configuration document of the API interface, parse the parameter configuration document, determine the basic configuration parameters of the API interface, and build an adaptation layer for different language algorithms based on the basic configuration parameters;

[0199] Obtain the basic information and algorithm files of different language algorithms entered by the user based on the user interface, and send the basic information and algorithm files of different language algorithms to the API interface;

[0200] The basic information and algorithm files of different language algorithms are verified based on the API interface, and after the verification, the algorithm files of different language algorithms are formatted based on the adaptation layer, and the formatted different language algorithms are cached.

[0201] The working principle and beneficial effects of the above technical solution are: by parsing the parameter configuration document of the API interface, the basic configuration parameters of the API interface can be accurately and effectively determined, so as to realize the accurate and effective construction of the adaptation layer of different language algorithms according to the basic configuration parameters, which provides convenience for the effective reception of different language algorithms. Secondly, the basic information and algorithm files of different language algorithms are received through the user interface, so as to facilitate the effective acquisition of the basic information and specific content of different language algorithms. Finally, the basic information and algorithm files of the received different language algorithms are verified, and the different language algorithms are formatted and cached after the verification, which provides convenience for the deployment and scheduling of different language algorithms.

[0202] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A multi-language compatible algorithm scheduling platform, characterized in that: include: The data collection module is used to receive basic information and algorithm files of different language algorithms based on the API interface, and convert the algorithm files into image files according to the basic information based on a preset method; The algorithm deployment module is used to configure the environment of the algorithm container based on basic information, and to deploy the image file in the algorithm container in isolation based on the environment configuration result, and to perform multi-dimensional management of the isolated and deployed algorithms of different languages; The algorithm scheduling module is used to receive business requirements based on multi-dimensional management results, lock the target language algorithm based on the business requirements, and schedule the target language algorithm based on the API interface.

2. A multi-language compatible algorithm scheduling platform according to claim 1, characterized in that: Data collection module, including: The interface configuration unit is used to obtain the parameter configuration document of the API interface, parse the parameter configuration document, determine the basic configuration parameters of the API interface, and build an adaptation layer for different language algorithms based on the basic configuration parameters; An information receiving unit, used to obtain basic information and algorithm files of different language algorithms entered by the user based on the user interface, and send the basic information and algorithm files of different language algorithms to the API interface; The data preprocessing unit is used to verify the basic information and algorithm files of different language algorithms based on the API interface, and format the algorithm files of different language algorithms based on the adaptation layer after the verification, and cache the formatted different language algorithms.

3. A multi-language compatible algorithm scheduling platform according to claim 2, characterized in that: Data preprocessing unit, including: The permission verification subunit is used to: Obtain data sources of different language algorithms based on the API interface, and determine the identity tags of device terminals corresponding to different language algorithms based on the data sources; The identity tag is matched with the preset permission library, and the upload permission of the device terminal corresponding to the different language algorithms is determined based on the data matching result. If the upload permission exists, the corresponding language algorithm is received based on the API interface, otherwise, the corresponding language algorithm is returned; The data verification sub-unit is used to verify the illegal characters and execution logic of different language algorithms based on the protocol texts of different language algorithms after the upload permission verification is passed, and input the different language algorithms into the adaptation layer for processing after the illegal characters and execution logic are verified.

4. The multi-language compatible algorithm scheduling platform according to claim 1, characterized in that: Data collection module, including: Image preparation unit, used to: Obtain basic information about different language algorithms, and match base images from the preset image library based on the basic information; At the same time, the operating requirements of different language algorithms are determined based on the basic information, and the corresponding necessary environment variables are determined based on the operating requirements; Image file generation unit, used for: Add algorithm files of different language algorithms to the base image, and configure the base image with the added algorithm files based on the necessary environment variables; Based on the configuration result, a preset image file generation mechanism is started, and based on the preset image file generation mechanism, the configuration result of the basic image with the algorithm file added is processed in a process-cooperative manner to obtain a final image file.

5. The multi-language compatible algorithm scheduling platform according to claim 1, characterized in that: Algorithm deployment module, including: Environment hive for: Determine the application scenarios of different language algorithms based on their basic information, and determine the runtime libraries and dependent software required by different language algorithms based on the application scenarios; Bind the runtime library and dependent software to the corresponding algorithm container, and complete the environment configuration of the algorithm container based on the binding results; Isolate deployment units for: Deploy different language algorithms in corresponding algorithm containers, and determine the current available resources and the target resources required for each algorithm container to run based on the deployment results; Adaptively allocate resource loads to each algorithm container based on available resources and target resources required for each algorithm container to run. At the same time, determine the importance parameters of different language algorithms based on basic information, and set container permissions for corresponding algorithm containers based on the adaptive allocation results and importance parameters to complete isolation deployment. Multi-dimensional management unit for: Based on the isolated deployment results, the update status of the corresponding language algorithm in each algorithm container is monitored in real time, and when there is an algorithm update, the algorithm update request and algorithm update file are received; Parse the algorithm update file based on the algorithm update request, determine the update parameters and the update nodes of the old version of the algorithm, and replace the parameters of the update nodes of the old version of the algorithm based on the update parameters; Generate an updated version of the algorithm based on the parameter replacement result, and authorize the updated version of the algorithm. At the same time, determine the data difference characteristics between the updated version of the algorithm and the old version of the algorithm, and determine the adaptive adjustment strategy for the algorithm container based on the data difference characteristics; The algorithm container is dynamically configured based on the adaptive adjustment strategy, and after the adjustment is completed, the old version of the algorithm is recorded and archived in sequence based on the update time node.

6. The multi-language compatible algorithm scheduling platform according to claim 5, characterized in that: Multi-dimensional management unit, including: A result acquisition subunit is used to obtain the isolation deployment results obtained, and determine the index tags corresponding to different algorithm containers based on the isolation deployment results; Management subunit, used to: Summarize the algorithm containers of different language algorithms to obtain algorithm container clusters, and build a search engine for the algorithm container clusters. At the same time, add index tags of different algorithm containers to the background script file of the search engine. Respond to the search request submitted by the user based on the added results, and locate the language algorithm to be managed based on the response results; An active management request submitted by a user to the managed language algorithm is obtained, and based on the active management request, a management mechanism is triggered to perform corresponding management operations on the managed language algorithm, wherein the active management request includes active update and deletion.

7. The multi-language compatible algorithm scheduling platform according to claim 1, characterized in that: Algorithm scheduling module, including: The demand receiving unit is used to receive the service demand sent by the user terminal and convert the service demand into text to obtain the demand data; A key data determination unit is used to perform semantic analysis on the demand data based on a neural network, obtain semantic features corresponding to the demand data, and determine key tasks and constraints of the business requirements based on the semantic features; An algorithm determination unit is used to extract the business features of different language algorithms, determine the subordination of key tasks and constraints, and lock the target language algorithm based on the subordination; The algorithm scheduling unit is used to schedule the target algorithm based on the API interface, and at the same time, synchronously schedule the resources required by the target algorithm.

8. The multi-language compatible algorithm scheduling platform according to claim 7, characterized in that: Algorithm scheduling unit, including: The condition setting subunit is used to set the preset trigger condition in the background and determine the associated factors included in the preset trigger condition; The automatic scheduling subunit is used to monitor the status of the associated factors in real time and trigger the scheduling operation of the required language algorithm when the status of the associated factors meets the preset trigger conditions; The recording subunit is used to collect the basic information and scheduling timestamp of the language algorithm corresponding to each scheduling operation based on the trigger result, and record the basic information and scheduling timestamp of the language algorithm corresponding to each scheduling operation, generate an algorithm scheduling log, and record the algorithm scheduling log.

9. A multi-language compatible algorithm scheduling method, characterized in that: include: Step 1: Receive basic information and algorithm files of different language algorithms based on the API interface, and convert the algorithm files into image files according to the basic information based on the preset method; Step 2: Configure the environment of the algorithm container based on the basic information, and deploy the image file in the algorithm container in isolation based on the environment configuration results, and manage the isolated and deployed different language algorithms in multiple dimensions; Step 3: Receive business requirements based on the multi-dimensional management results, lock the target language algorithm based on the business requirements, and schedule the target language algorithm based on the API interface.

10. The multi-language compatible algorithm scheduling method according to claim 9, characterized in that: Step 1: Receive basic information and algorithm files of different language algorithms based on the API interface, including: Obtain the parameter configuration document of the API interface, parse the parameter configuration document, determine the basic configuration parameters of the API interface, and build an adaptation layer for different language algorithms based on the basic configuration parameters; Obtain the basic information and algorithm files of different language algorithms entered by the user based on the user interface, and send the basic information and algorithm files of different language algorithms to the API interface; The basic information and algorithm files of different language algorithms are verified based on the API interface, and after the verification, the algorithm files of different language algorithms are formatted based on the adaptation layer, and the formatted different language algorithms are cached.

Images