Artificial intelligence service generation method, device and computer-readable storage medium

Abstract

The present invention proposes an artificial intelligence business generation method, device and computer-readable storage medium, wherein the artificial intelligence business generation method includes: obtaining a preset general algorithm model; determining a business data set corresponding to the general algorithm model; determining training data for training in the business data set; generating evaluation data according to the training data and the preset general algorithm model; receiving generation instructions corresponding to the evaluation data; generating artificial intelligence business according to the generation instructions. According to the technical solution of the present invention, the user does not need to edit the code during the whole process, the model training is simple, and the professional requirements of the user are greatly reduced. At the same time, the preset general algorithm model is used to greatly improve the reusability and versatility of the algorithm model.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence business generation technology, and in particular to an artificial intelligence business generation method, an artificial intelligence business generation device and a computer-readable storage medium. Background Art

[0002] The threshold for existing artificial intelligence algorithms is high, requiring professional requirements such as a certain mathematical foundation and code coding ability, and the versatility of algorithms in this field is low. For AI models, developers are required to perform professional development operations from creation, training to online launch. Especially in similar application fields, there will be more repetitive model building processes, which greatly increases the workload of developers. Summary of the invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art or related art.

[0004] In view of this, a first aspect of an embodiment of the present invention provides an artificial intelligence service generation method.

[0005] A second aspect of an embodiment of the present invention provides an artificial intelligence service generation device.

[0006] A third aspect of an embodiment of the present invention provides a computer-readable storage medium.

[0007] In order to achieve the above-mentioned objectives, an embodiment of the first aspect of the present invention provides an artificial intelligence business generation method, including: obtaining a preset general algorithm model; determining a business data set corresponding to the general algorithm model; determining training data for training in the business data set; generating evaluation data based on the training data and the preset general algorithm model; receiving generation instructions corresponding to the evaluation data; and generating an artificial intelligence business according to the generation instructions.

[0008] According to the artificial intelligence business generation method proposed in the present invention, the preset general algorithm model set in advance is first obtained, and the corresponding business data set is obtained, and then the corresponding training data is determined, and the training data can be used to train the above model. In actual use, when using artificial intelligence business in Shanghai, users do not need professional algorithm model development technology, that is, they do not need professional code coding ability, and can achieve the generation of targeted business. Specifically, after determining the corresponding business data set, it is necessary to determine the training data for training the model in the business data set. It can be understood that the training data is generally data stored in advance, generally including a training set and a verification set. The data in the training set is first imported into the preset general algorithm model, and the obtained data is then compared with the data corresponding to the verification set, and the specific training results can be displayed. Under different evaluation criteria, different evaluation data can be generated, so that users can judge the training effect according to specific evaluation data, thereby obtaining instructions on whether to use the model. When the user believes that the training effect is good after training with the training data in the current business data set, a generation instruction will be issued. At this time, it is believed that the preset general algorithm model that has been trained in a targeted manner can meet the user's usage needs, so the corresponding artificial intelligence business can be generated directly according to the received generation instruction.

[0009] During the entire process, users do not need to edit any code, and model training is simple, which greatly reduces the professional requirements for users. At the same time, the use of preset general algorithm models greatly improves the reusability and versatility of the algorithm model.

[0010] It can be understood that the preset general algorithm model is a model developed in advance by the developer, and the user only needs to migrate or retrain the model based on the model to obtain services suitable for the specific environment. In other words, in this embodiment, the developed algorithm model can be used to provide targeted intelligent services. In particular, for users with the same needs, the developer can develop a more general algorithm model, and different users can download the same preset general algorithm model and train it according to different training data, so as to obtain artificial intelligence services that are more suitable for different user usage scenarios.

[0011] In addition, the artificial intelligence service generation method in the above solution provided by the present invention may also have the following additional technical features:

[0012] In the above technical solution, determining the business data set corresponding to the general algorithm model specifically includes: registering the business data set in the data management system; and importing metadata corresponding to the general algorithm model into the business data set.

[0013] In this technical solution, when the business data set is determined, it is mainly obtained by registering in a unified data management system. After registration, metadata can be imported into the business data set. The metadata can be pre-annotated or completely unannotated data. By importing metadata into the business data set, it is convenient for subsequent screening of training data, thereby facilitating model training.

[0014] It is understandable that users can register annotated or unannotated data in a unified way in data management to form the required data set. There are two main ways to register a data set: one is to create a new data set, and the other is to use an external data set. When the data set is an external data set, it can be registered by importing files, or by using data from a third-party annotation platform.

[0015] In the above technical solution, before determining the business data set corresponding to the general algorithm model, it also includes: obtaining a domain data lake; receiving a synchronization instruction; and determining metadata in the domain data lake according to the synchronization instruction.

[0016] In this technical solution, a new step of data domain synchronization is added. Specifically, before determining the business data set, it is necessary to obtain the domain data lake in order to facilitate the subsequent determination of metadata for a specific domain. By receiving the synchronization instruction, the data in the domain data lake can be screened according to certain conditions, and the data that does not meet the current preset general algorithm model can be eliminated. The screened data can be synchronized, that is, part of the data in the domain data lake is determined as metadata for model training, which can improve the evaluation data after model training and make it more satisfactory for the use needs of professional fields.

[0017] It should be noted that a domain data lake is a large-scale data repository and processing engine for a specific domain, which can store a large amount of data in a specific domain.

[0018] In the above technical solution, obtaining a preset general algorithm model specifically includes: obtaining a code framework; obtaining an algorithm model corresponding to the code framework uploaded to the business platform, wherein the algorithm model can be migrated or retrained according to the application scenario.

[0019] In this technical solution, when obtaining the preset general algorithm model, the code framework is first obtained. After the developer develops based on the code framework, the algorithm model can be completed. When the algorithm model is uploaded to the business platform, different users can download the pre-developed algorithm model from the business platform, thereby facilitating subsequent targeted training.

[0020] It should be noted that the developed algorithm model can be migrated according to similar application scenarios, or it can be retrained according to application scenarios with slight differences.

[0021] Those skilled in the art should understand that the code framework is a scaffolding framework. Developers can export the code framework through a specific platform, that is, a business platform, and then use specific development tools to develop algorithm models. It is understandable that developers have a certain mathematical foundation and code coding ability, which is different from the business requirements of specific users.

[0022] Among them, development tools include but are not limited to local integrated development tools (ie, IDE) and online inheritance tools, such as nodebook.

[0023] The above technical solution also includes: receiving an inference request corresponding to an artificial intelligence business; generating an external access address based on the inference request; or performing real-time inference based on the inference request.

[0024] In this technical solution, when there is a need for reasoning, it is mainly achieved by whether an reasoning request is received. When an reasoning request corresponding to an artificial intelligence business is received, there are two reasoning methods. One is to directly generate an external access address, thereby generating reasoning data for different links in actual application scenarios, or it can be directly registered in a specific form object. The list calls the intelligent service through the UI metadata value bound to the UI template to perform online real-time reasoning.

[0025] The application scenarios of inference data include but are not limited to key links such as marketing, procurement, human resource management, supply chain, finance, and manufacturing.

[0026] In the above technical solution, determining the training data for training in the business data set specifically includes: determining the training set and the verification set in the business data set; determining the ratio of the training set and the verification set; and determining the training data according to the ratio.

[0027] In this technical solution, when determining the training data, the training set and validation set in the business data set are first screened and determined, and the ratio of the two is determined, so as to determine the corresponding training data according to different ratios. It can be understood that the training set and validation set are both learning samples in the machine learning process. The training set is mainly used to establish the model and is a sample set mainly used for learning. The validation set is used to determine the parameters of the network structure or control the complexity of the model. In other words, the validation set can be used to determine whether the algorithm model meets the user's expectations.

[0028] Among them, the different ratios of the training set and the validation set will have a certain impact on the training data, so it is necessary to train the algorithm model based on the determined ratio of the two.

[0029] The above technical solution also includes: obtaining a selection instruction for a domain application tree; determining metadata corresponding to the domain application tree according to the selection instruction; adjusting the ratio of the training set and the validation set; and determining training data according to the adjusted ratio.

[0030] In this technical solution, feature processing is mainly performed on data from different fields. Specifically, the corresponding field application tree is first selected, that is, the selection instruction corresponding to the specific field application tree is obtained. According to the selection instruction, the corresponding metadata will be determined. On this basis, the training data can be fine-tuned by adjusting the ratio of the training set and the verification set, so as to facilitate feature processing based on the enterprise metadata and the entity data of the metadata, so as to complete targeted adjustments.

[0031] An embodiment of the second aspect of the present invention provides an artificial intelligence business generation device, including: a memory, in which a computer program is stored; a processor is used to implement the steps of any artificial intelligence business generation method in the embodiment of the first aspect when executing the computer program.

[0032] According to an embodiment of the artificial intelligence business generation device of the present invention, it includes a processor and a memory. Since the processor can execute a computer program or instruction stored in the memory and implement any one of the artificial intelligence business generation methods of the first aspect when executing the computer program or instruction, the artificial intelligence business generation device of the present invention has all the beneficial effects of the artificial intelligence business generation method in any one of the above technical solutions, which will not be repeated here.

[0033] An embodiment of the third aspect of the present invention provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the artificial intelligence business generation method are implemented.

[0034] Through the embodiment of the computer-readable storage medium of the present invention, a computer program is stored thereon, and when the computer program is executed by the processor, the steps of the artificial intelligence business generation method in any of the above embodiments are implemented, thereby having all the beneficial effects of the artificial intelligence business generation method in any of the above embodiments, which will not be repeated here.

[0035] Additional aspects and advantages of the present invention will become apparent from the following description or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A schematic diagram showing a process flow of an artificial intelligence service generation method according to an embodiment of the present invention is shown;

[0037] Figure 2A schematic diagram showing a process flow of an artificial intelligence service generation method according to an embodiment of the present invention;

[0038] Figure 3 A schematic diagram showing a process flow of an artificial intelligence service generation method according to an embodiment of the present invention is shown;

[0039] Figure 4 A schematic diagram showing a process flow of an artificial intelligence service generation method according to an embodiment of the present invention is shown;

[0040] Figure 5 A schematic block diagram of the structure of an artificial intelligence service generation device according to an embodiment of the present invention is shown.

[0041] in, Figure 5 The corresponding relationship between the reference numerals and the component names is as follows:

[0042] 100: artificial intelligence service generation device; 102: memory; 104: processor. DETAILED DESCRIPTION

[0043] In order to more clearly understand the above-mentioned purposes, features and advantages of the embodiments of the present invention, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0044] In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the embodiments of the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present application is not limited to the specific embodiments disclosed below.

[0045] Refer to the following Figures 1 to 5 Some embodiments according to the invention are described.

[0046] Embodiment 1:

[0047] like Figure 1 As shown, in one embodiment of the present invention, a method for generating an artificial intelligence business is provided, including: step S1002, obtaining a preset general algorithm model; step S1004, determining a business data set corresponding to the general algorithm model; step S1006, determining training data for training in the business data set; step S1008, generating evaluation data based on the training data and the preset general algorithm model; step S1010, receiving a generation instruction corresponding to the evaluation data; step S1012, generating an artificial intelligence business according to the generation instruction.

[0048] First, the preset general algorithm model set in advance is obtained, and the corresponding business data set is obtained, and then the corresponding training data is determined, and the training data can be used to train the above model. In actual use, when using artificial intelligence business in Shanghai, users do not need professional algorithm model development technology, that is, they do not need professional code coding ability to achieve the generation of targeted business. Specifically, after determining the corresponding business data set, it is necessary to determine the training data for training the model in the business data set. It can be understood that the training data is generally data stored in advance, generally including a training set and a verification set. First, the data in the training set is imported into the preset general algorithm model, and the obtained data is then compared with the data corresponding to the verification set to display the specific training results. Under different evaluation criteria, different evaluation data can be generated, so that users can judge the training effect according to specific evaluation data, so as to obtain instructions on whether to use the model. When the user believes that the training effect is good after training with the training data in the current business data set, a generation instruction will be issued. At this time, it is believed that the preset general algorithm model that has been trained in a targeted manner can meet the user's usage needs, so the corresponding artificial intelligence business can be generated directly according to the received generation instruction.

[0049] During the entire process, users do not need to edit any code, and model training is simple, which greatly reduces the professional requirements for users. At the same time, the use of preset general algorithm models greatly improves the reusability and versatility of the algorithm model.

[0050] It can be understood that the preset general algorithm model is a model developed in advance by the developer, and the user only needs to migrate or retrain the model based on the model to obtain services suitable for the specific environment. In other words, in this embodiment, the developed algorithm model can be used to provide targeted intelligent services. In particular, for users with the same needs, the developer can develop a more general algorithm model, and different users can download the same preset general algorithm model and train it according to different training data, so as to obtain artificial intelligence services that are more suitable for different user usage scenarios.

[0051] Embodiment 2:

[0052] like Figure 2As shown, in one embodiment of the present invention, a method for generating an artificial intelligence business is provided, including: step S2002, obtaining a preset general algorithm model; step S2004, registering a business data set in a data management system; step S2006, importing metadata corresponding to the general algorithm model in the business data set; step S2008, determining training data for training in the business data set; step S2010, generating evaluation data based on the training data and the preset general algorithm model; step S2012, receiving a generation instruction corresponding to the evaluation data; step S2014, generating an artificial intelligence business according to the generation instruction.

[0053] When determining the business data set, it is mainly obtained by registering in a unified data management system. After registration, metadata can be imported into the business data set. The metadata can be pre-labeled or completely unlabeled data. By importing metadata into the business data set, it is convenient for subsequent screening of training data, thereby facilitating model training.

[0054] It is understandable that users can register annotated or unannotated data in a unified way in data management to form the required data set. There are two main ways to register a data set: one is to create a new data set, and the other is to use an external data set. When the data set is an external data set, it can be registered by importing files, or by using data from a third-party annotation platform.

[0055] Embodiment three:

[0056] like Figure 3 As shown, in one embodiment of the present invention, a method for generating an artificial intelligence business is provided, including: step S3002, obtaining a preset general algorithm model; step S3004, obtaining a domain data lake; step S3006, receiving a synchronization instruction; step S3008, determining metadata in the domain data lake according to the synchronization instruction; step S3010, registering a business data set in a data management system; step S3012, importing metadata corresponding to the general algorithm model in the business data set; step S3014, determining training data for training in the business data set; step S3016, generating evaluation data according to the training data and the preset general algorithm model; step S3018, receiving a generation instruction corresponding to the evaluation data; step S3020, generating an artificial intelligence business according to the generation instruction.

[0057] In this technical solution, a new step of data domain synchronization is added. Specifically, before determining the business data set, it is necessary to obtain the domain data lake in order to facilitate the subsequent determination of metadata for a specific domain. By receiving the synchronization instruction, the data in the domain data lake can be screened according to certain conditions, and the data that does not meet the current preset general algorithm model can be eliminated. The screened data can be synchronized, that is, part of the data in the domain data lake is determined as metadata for model training, which can improve the evaluation data after model training and make it more satisfactory for the use needs of professional fields.

[0058] It should be noted that a domain data lake is a large-scale data repository and processing engine for a specific domain, which can store a large amount of data in a specific domain.

[0059] Embodiment 4:

[0060] like Figure 4 As shown, in one embodiment of the present invention, a method for generating an artificial intelligence business is provided, including: step S4002, obtaining a code framework; step S4004, obtaining an algorithm model corresponding to the code framework uploaded to a business platform; step S4006, determining a business data set corresponding to the general algorithm model; step S4008, determining training data for training in the business data set; step S4010, generating evaluation data based on the training data and a preset general algorithm model; step S4012, receiving a generation instruction corresponding to the evaluation data; step S4014, generating an artificial intelligence business according to the generation instruction.

[0061] When obtaining the preset general algorithm model, the code framework is first obtained. After the developer develops based on the code framework, the algorithm model can be completed. When the algorithm model is uploaded to the business platform, different users can download the pre-developed algorithm model from the business platform, thereby facilitating subsequent targeted training.

[0062] It should be noted that the developed algorithm model can be migrated according to similar application scenarios, or it can be retrained according to application scenarios with slight differences.

[0063] Those skilled in the art should understand that the code framework is a scaffolding framework. Developers can export the code framework through a specific platform, that is, a business platform, and then use specific development tools to develop algorithm models. It is understandable that developers have a certain mathematical foundation and code coding ability, which is different from the business requirements of specific users.

[0064] Among them, development tools include but are not limited to local integrated development tools (ie, IDE) and online inheritance tools, such as nodebook.

[0065] Based on any of the above embodiments, it also includes: receiving an inference request corresponding to an artificial intelligence business; generating an external access address according to the inference request; or performing real-time inference according to the inference request.

[0066] When there is a need for reasoning, it is mainly realized by whether an reasoning request is received. When an reasoning request corresponding to an artificial intelligence business is received, there are two reasoning methods. One is to directly generate an external access address, thereby generating reasoning data for different links in actual application scenarios, or it can be directly registered in a specific form object. The list calls the intelligent service through the UI metadata value bound to the UI template for online real-time reasoning.

[0067] The application scenarios of inference data include but are not limited to key links such as marketing, procurement, human resource management, supply chain, finance, and manufacturing.

[0068] Based on any of the above embodiments, determining training data for training in the business data set specifically includes: determining a training set and a validation set in the business data set; determining a ratio of the training set to the validation set; and determining the training data according to the ratio.

[0069] When determining the training data, the training set and validation set in the business data set are mainly screened and determined first, and the ratio of the two is determined, so as to determine the corresponding training data according to different ratios. It can be understood that the training set and validation set are both learning samples in the machine learning process. The training set is mainly used to build the model and is the sample set mainly used for learning. The validation set is used to determine the parameters of the network structure or control the complexity of the model. In other words, the validation set can be used to determine whether the algorithm model meets the user's expectations.

[0070] Among them, the different ratios of the training set and the validation set will have a certain impact on the training data, so it is necessary to train the algorithm model based on the determined ratio of the two.

[0071] In addition, based on any of the above embodiments, it also includes: obtaining a selection instruction for a domain application tree; determining metadata corresponding to the domain application tree according to the selection instruction; adjusting the ratio of the training set and the validation set; and determining training data according to the adjusted ratio.

[0072] This embodiment mainly performs feature processing on data in different fields. Specifically, the corresponding field application tree is first selected, that is, the selection instruction corresponding to the specific field application tree is obtained. According to the selection instruction, the corresponding metadata will be determined. On this basis, the training data can be fine-tuned by adjusting the ratio of the training set and the verification set, so as to facilitate feature processing based on the enterprise metadata and the entity data of the metadata, so as to complete the targeted adjustment.

[0073] Embodiment five:

[0074] like Figure 5 As shown, an embodiment of the present invention provides an artificial intelligence service generation device 100. The artificial intelligence service generation device 100 includes a memory 102; a processor 104. Since the processor can execute a computer program or instruction stored in the memory and implement any of the above artificial intelligence service generation methods when executing the computer program or instruction, the artificial intelligence service generation device has all the beneficial effects of the artificial intelligence service generation method in any of the above embodiments, which will not be repeated here.

[0075] Embodiment six:

[0076] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the control method in any of the above embodiments can be implemented.

[0077] In a specific embodiment, a codeless AI business generator is mainly disclosed. It is oriented to domain customers and provides visual operation interface services based on preset and accumulated intelligent algorithms and models to quickly respond to the complex and changeable front-end business. It provides a visual, wizard-based interactive interface to support reasoning data covering key links required for marketing, procurement, OA, HR management, supply chain, finance, manufacturing, etc. involved in business services of enterprises, including prediction, recognition, classification, decision-making and other typical scenarios, and quickly realizes intelligent services with one click. The domain can quickly and conveniently configure training scenarios based on existing data visualization to build algorithm models for intelligent services.

[0078] This embodiment is mainly aimed at the intelligentization of enterprise-level AI applications, and realizes self-training through pre-made algorithm models, enterprise-configured feature parameters and its own metadata data sets, omitting the process of AI model creation, training and deployment that requires developers to develop and operate. Domain business personnel can achieve code-free development through page visualization operations. When serving the needs of intelligent scenarios in the service field, simple algorithms and basic functions lack product services close to the business, forming a service fault. Therefore, on top of building basic AI capabilities, the intelligent platform quickly precipitates the intelligent algorithms required by the business, and directly builds general intelligent application products to support domain needs in the form of visual components or data services.

[0079] This embodiment can achieve the following effects: using the algorithms that have been developed and verified to preset the platform to provide enterprise intelligent services; allowing non-algorithm developers to quickly realize intelligent service scenarios and build reasoning services; quickly binding the business data generated in the field to the model training operation through metadata or data source configuration is simple and code-free; solving the generalization ability of machine learning products, so that it can realize the reuse of algorithm models in similar business scenarios in more fields.

[0080] The code-free AI platform consists of preset algorithms, business objects, business types, data source configuration, data preprocessing, algorithm models, evaluation and reporting, training services, and reasoning services, which can quickly realize intelligent service scenarios for business processes.

[0081] Specifically, for preset general algorithms, developers export the scaffolding framework through the platform, complete the algorithm model development through the local IDE or the online integrated tool nodebook, and upload the preset to the platform for business personnel to migrate and retrain models according to similar scenarios to meet the needs and achieve intelligence. The algorithm library installs and renders the algorithm according to the algorithm metadata information and the algorithm installation package (including algorithm metadata, algorithm installation package and installation program, algorithm entry packaging program), and manages the successfully rendered algorithms.

[0082] For data sets, the data in the domain lake is used to create data sets, and the data model service data synchronization function is used to realize the physicalization of the data model, which can bind the metadata information description; users can manage the labeled or unlabeled data in a unified way in the data management to form the required data set. There are two ways to register a data set: create a new data set and an external data set (import CSV files, third-party labeling platform labelme).

[0083] As for the training effect, by selecting the data model or metadata information for training, the evaluation data such as AUC / ROC / MSE related to the model training results are displayed. Users can evaluate whether the model is usable based on the effect.

[0084] For the release of model training business objects, the published inference service can generate an API external access address and request examples; it can also be directly registered in the form object, and the list calls the intelligent service through the UI metadata value bound to the UI template to perform online real-time inference; the front end identifies the UI metadata information generated by the intelligent service. If it is the UI metadata of the intelligent service, the front end requests a fixed interface and transmits form-related data to asynchronously bind the form data. The inference of the intelligent service returns the result set through a high-performance inference engine, and parameter conversion is used for the child or grandchild table information that the front-end UI form cannot obtain. After conversion, the intelligent service calls the corresponding data to match the inference parameters to obtain the inference result.

[0085] For AI model metadata, it is aimed at specific scenarios and specific algorithms, and is a combination of clear algorithm metadata, feature model metadata, and data model metadata. (That is, AI model metadata = algorithm metadata + feature model metadata + data model metadata)

[0086] The scheme proposed for the above specific embodiments can achieve the following specific effects: Support simple applications for non-professional roles such as enterprise operations, business, sales, and network management. The code-free, wizard-style interface creates and launches the model process, which realizes the connection of the AI ​​development process. The independently developed intelligent algorithm is integrated into the platform, which connects the entire process of AI model creation, training, and deployment, and is transparent and imperceptible to the field. When the product capabilities provided by the infrastructure meet the needs of intelligent scenarios in the field, the simple algorithms and basic functions lack product services close to the business, forming a service gap. Therefore, on top of building basic AI capabilities, the code-free intelligent platform quickly precipitates the intelligent algorithms required by the business, and directly builds general intelligent application products to support field needs in the form of visual components or data services.

[0087] Based on the enterprise metadata and metadata entity data feature processing, select the domain application tree, adjust the ratio of training set and verification set through the metadata selected by the corresponding domain application tree, and publish the reasoning to add pseudo columns to the UI metadata to achieve the form business data display.

[0088] Enterprises can pre-set their own developed algorithms, and the platform will also have available general algorithm models that have been verified, allowing for easy out-of-the-box use and operational process guidance.

[0089] Taking the scenario as an example, the intelligent demand in the field is to provide customers with intelligent services in actual production scenarios in the field products based on business status. In the marketing and retail scenario, physical stores need to make effective predictions on future product sales in the aspects of replenishment, pricing, promotional activity strategy, inventory management, etc., in order to ensure sales while controlling costs, improving turnover rate, reducing loss rate, and rationally allocating inventory to ultimately maximize profits. Intelligent services conduct statistical analysis on store product sales and provide customers with forecasting services based on product sales in combination with forecasting algorithms. They support enhancing forecasting effects through rule configuration, and can view forecasting effects in real time through promotion, pricing, and other adjustments, as well as the impact on inventory turnover, profit margin, and loss rate, to assist in decision-making for store operations.

[0090] Partial feature engineering automation and data labeling automation capabilities are aimed at enterprise operation scenarios such as marketing, procurement, office, HR management, supply chain, finance, and manufacturing. Based on the model ontology-parameter separation architecture, for specific scenarios, the integration of business knowledge and machine learning is achieved by calling the pre-trained feature engineering process or the pre-defined data labeling process.

[0091] Through preset algorithms such as prediction generator, decision generator, general OCR recognition generator, etc., we can achieve complete intelligent solutions based on industry or field characteristics, including smart marketing and smart procurement.

[0092] The above describes in detail the technical solution of the present invention in combination with the accompanying drawings. During the entire process, users do not need to edit the code, and the model training is simple, which greatly reduces the professional requirements of users. At the same time, the preset general algorithm model is used to greatly improve the reusability and versatility of the algorithm model.

[0093] In the present invention, the terms "first", "second", and "third" are used for descriptive purposes only and should not be understood as indicating or implying relative importance; the term "plurality" refers to two or more, unless otherwise clearly defined. The terms "installed", "connected", "connected", "fixed", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; "connected" can be a direct connection or an indirect connection through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0094] In the description of the present invention, it is necessary to understand that the directions or positional relationships indicated by terms such as “upper”, “lower”, “left”, “right”, “front” and “back” are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or unit referred to must have a specific direction, be constructed and operated in a specific orientation, and therefore, cannot be understood as a limitation on the present invention.

[0095] In the description of this specification, the description of the terms "one embodiment", "some embodiments", "specific embodiments", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0096] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for generating artificial intelligence services, characterized in that: include: Obtain the preset general algorithm model; Determine a business data set corresponding to the general algorithm model; Determining training data for training in the business data set; Generate evaluation data according to the training data and the preset general algorithm model; receiving a generation instruction corresponding to the evaluation data; Generate an artificial intelligence service according to the generation instruction; The determining of the business data set corresponding to the general algorithm model specifically includes: Registering the business data set in a data management system; Importing metadata corresponding to the general algorithm model into the business data set; Before determining the business data set corresponding to the general algorithm model, the method further includes: Get the domain data lake; Receive synchronization instructions; determining the metadata in the domain data lake according to the synchronization instruction; The obtaining of the preset general algorithm model specifically includes: Get the code framework; Obtain the algorithm model corresponding to the code framework uploaded to the business platform, The algorithm model can be migrated or retrained according to the application scenario; The determining of the training data for training in the business data set specifically includes: Determine a training set and a validation set in the business data set; Determining the ratio of the training set to the validation set; Determine the training data according to the ratio; Among them, the training set and the verification set are both learning samples in the machine learning process. The training set is mainly used to establish a model and is a sample set mainly used for learning, while the verification set is used to determine the network structure or control the parameters of the model complexity.

2. The artificial intelligence service generation method according to claim 1, characterized in that: Also includes: Receiving an inference request corresponding to the artificial intelligence service; Generate an external access address according to the inference request; or Real-time reasoning is performed according to the reasoning request.

3. The artificial intelligence service generation method according to claim 1, characterized in that: Also includes: Get the selection instruction of the domain application tree; Determining metadata corresponding to the domain application tree according to the selection instruction; Adjusting the ratio of the training set to the validation set; The training data is determined according to the adjusted ratio.

4. An artificial intelligence service generation device, characterized in that: include: a memory, wherein a computer program is stored in the memory; A processor, wherein the processor is used to implement the steps of the artificial intelligence business generation method as described in any one of claims 1 to 3 when executing the computer program.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the artificial intelligence business generation method according to any one of claims 1 to 3 are implemented.

Images