Cloud service recommendation method and apparatus
By training a deep learning model to predict the processing time of cloud service resources, the problem of excessively long processing time in cloud service recommendations is solved, achieving the effects of accurate recommendations and fast processing.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA MOBILE INFORMATION TECHNOLOGY CO LTD
- Filing Date
- 2022-01-04
- Publication Date
- 2026-06-19
Smart Images
Figure CN116415063B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of cloud service technology, specifically to a cloud service recommendation method and apparatus. Background Technology
[0002] Cloud service platforms offer a wide variety of services. To recommend suitable cloud services to users, one related technology utilizes a MapReduce parallel computing model to efficiently mine frequent itemsets in a distributed system. Using these frequent itemsets as samples, a deep learning network is built and trained, which then provides users with accurate service resource recommendations.
[0003] However, depending on the service delivery method, the service acceptance process and service acceptance time often vary greatly. The relevant technologies do not take into account the service acceptance time, which may result in the service resources recommended to users taking too long to be accepted, thus affecting the user experience. Summary of the Invention
[0004] This application provides a cloud service recommendation method and apparatus that can recommend accurate service resources to users while optimizing the processing time of the recommended service resources.
[0005] In a first aspect, embodiments of this application provide a cloud service recommendation method, including:
[0006] The current user information is input into the trained first deep learning model, so that, based on the first deep learning model, at least one target resource frequent item that matches the current user information is determined from the frequent item set formed by the service resources of historical orders.
[0007] Input each frequent item of the target resource into the trained second deep learning model to determine the service acceptance time of each frequent item of the target resource.
[0008] Based on the service acceptance time of each of the target resource frequent items, the target resource frequent item with the shortest service acceptance time is selected for service recommendation;
[0009] The first deep learning model is trained using historical customer information from the historical orders and the set of frequent items. The set of frequent items includes multiple frequent resource items, and the frequent resource items include multiple service resources.
[0010] The second deep learning model is trained using the historical customer information, the set of frequent items, and the service processing time of each service resource in the historical orders.
[0011] In one embodiment, it also includes:
[0012] Based on the Apriori algorithm, frequent items are extracted from the resource types of each service resource in historical orders to obtain frequent items of each type;
[0013] Based on the frequent items of each category, extract the set of resource model combinations corresponding to each frequent item of each category from the historical orders;
[0014] Based on the frequency of occurrence of any resource model combination in the target resource model combination set in each resource model combination set, extract the frequent resource model items whose frequency of occurrence is greater than a preset threshold from the target resource model combination set.
[0015] The frequent items of resources are determined based on the frequent items of the model and the frequent items of the type corresponding to the combination set of the target resource models;
[0016] The first deep learning model is built and trained based on the set of frequent resource items and historical customer information.
[0017] The resource model combination includes multiple resource models.
[0018] In one embodiment, the resource type includes at least one of storage, memory, and CPU.
[0019] In one embodiment, it also includes:
[0020] The service processing time of each of the target resources with frequent occurrences is pushed to the user terminal corresponding to the current user information.
[0021] In one embodiment, the frequency of the frequently occurring resource item in the historical orders is greater than a preset frequency.
[0022] In one embodiment, the frequent resource items include at least a set of coupled resources;
[0023] The coupled resources are the service resources that must be applied for simultaneously when applying for cloud service resource onboarding.
[0024] In one embodiment, the service resources in the frequent resource items are of different types.
[0025] Secondly, embodiments of this application provide a cloud service recommendation device, comprising:
[0026] The frequent item determination module is used to input the current user information into a trained first deep learning model, so as to determine at least one target resource frequent item that matches the current user information from the frequent item set formed by the service resources of historical orders according to the first deep learning model;
[0027] The duration determination module is used to input the frequent items of each target resource into the trained second deep learning model to determine the service acceptance duration of each frequent item of the target resource.
[0028] The service recommendation module is used to recommend services based on the service acceptance time of each of the target resource frequent items, and to obtain the target resource frequent item with the shortest service acceptance time.
[0029] The first deep learning model is trained using historical customer information from the historical orders and the set of frequent items. The set of frequent items includes multiple frequent resource items, and the frequent resource items include multiple service resources.
[0030] The second deep learning model is trained using the historical customer information, the set of frequent items, and the service processing time of each service resource in the historical orders.
[0031] Thirdly, embodiments of this application provide an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the program to implement the steps of the cloud service recommendation method described in the first aspect.
[0032] Fourthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the steps of the cloud service recommendation method described in the first aspect.
[0033] The cloud service recommendation method and apparatus provided in this application embodiment uses a first deep learning model trained on a set of frequent service resources as samples to match multiple frequent target resource items corresponding to the current user information. Then, a second deep learning model trained on the service processing time of the service resources selects the frequent target resource item with the shortest processing time from the multiple frequent target resource items matched with the current user information for service recommendation. This approach considers both the accuracy of the recommended service resources and their processing time, thereby recommending accurate service resources to the user while optimizing the processing time of the recommended service resources based on the second deep learning model. Furthermore, it eliminates the need for users to individually select different service resources, reducing the user's application time. Attached Figure Description
[0034] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0035] Figure 1This is a flowchart illustrating the cloud service recommendation method provided in an embodiment of the present invention;
[0036] Figure 2 This is a schematic diagram illustrating the extraction of frequently occurring items according to an embodiment of the present invention;
[0037] Figure 3 This is a schematic diagram illustrating the extraction of frequently used model items provided in an embodiment of the present invention;
[0038] Figure 4 This is a schematic diagram illustrating the extraction of frequently occurring resource items provided in an embodiment of the present invention;
[0039] Figure 5 This is a schematic diagram of the cloud service recommendation device provided by the present invention;
[0040] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation
[0041] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0042] To better understand the solution, the technical terms involved in the embodiments of the present invention are explained as follows:
[0043] Resource Coupling: When applying for cloud service resources, it is common to encounter situations where two or more resources must be applied for simultaneously. For example, when applying for visualization and analysis data services, storage, memory, and CPU resources must be selected at the same time. This characteristic of cloud service resources that must be applied for simultaneously is called resource coupling, and the corresponding service resources are called coupled resources.
[0044] Model Exclusivity: When applying for onboarding, only one model of each cloud service resource can be selected. For example, memory resources can only be selected from multiple models such as 2G, 4G, 8G, and 16G. This property is called model exclusivity.
[0045] The cloud service recommendation method provided in this application will be described in detail and explained below through several specific embodiments.
[0046] like Figure 1 As shown, in one embodiment, a cloud service recommendation method is provided. This embodiment mainly illustrates the application of this method to computer devices or servers.
[0047] Reference Figure 1 The cloud service recommendation method provided in this embodiment includes:
[0048] Step 101: Input the current user information into the trained first deep learning model, so as to determine at least one target resource frequent item that matches the current user information from the frequent item set formed by the service resources of historical orders according to the first deep learning model;
[0049] Step 102: Input each frequent item of the target resource into the trained second deep learning model to determine the service acceptance time of each frequent item of the target resource;
[0050] Step 103: Based on the service acceptance time of each of the target resource frequent items, obtain the target resource frequent item with the shortest service acceptance time and recommend services accordingly;
[0051] The first deep learning model is trained using historical customer information from the historical orders and the set of frequent items. The set of frequent items includes multiple frequent resource items, and the frequent resource items include multiple service resources.
[0052] The second deep learning model is trained using the historical customer information, the set of frequent items, and the service processing time of each service resource in the historical orders.
[0053] A first deep learning model, trained on a set of frequent service resource items, matches multiple frequent target resource items to the current user information. A second deep learning model, trained on the service processing time of each service resource, selects the frequent target resource with the shortest processing time from the matching items. This approach balances the accuracy of the recommended service resources with their processing time, optimizing the processing time of the recommended services while providing precise recommendations. Furthermore, it eliminates the need for users to individually select different service resources, reducing application time.
[0054] In one embodiment, the user information includes the applicant's name, the application field of the cloud service applied for, the type of cloud service applied for, and the applicant's number.
[0055] In one embodiment, historical orders include multiple historical order data entries. Each historical order data entry includes historical customer information, the service resources requested by the historical customer information, and the processing time of the historical order data entry, i.e., the processing time of each service resource in the historical order data entry.
[0056] In one embodiment, the training method for the first deep learning model includes:
[0057] Based on the Apriori algorithm, frequent items are extracted from the resource types of each service resource in historical orders to obtain frequent items of each type;
[0058] Based on the frequent items of each category, extract the set of resource model combinations corresponding to each frequent item of each category from the historical orders;
[0059] Based on the frequency of occurrence of any resource model combination in the target resource model combination set in each resource model combination set, extract the frequent resource model items whose frequency of occurrence is greater than a preset threshold from the target resource model combination set.
[0060] The frequent items of resources are determined based on the frequent items of the model and the frequent items of the type corresponding to the combination set of the target resource models;
[0061] The first deep learning model is built and trained based on the set of frequent resource items and historical customer information.
[0062] The resource model combination includes multiple resource models.
[0063] In one embodiment, assuming the resource type set of each service resource requested by historical customer information, such as storage, memory, CPU, etc., is I = {1, 2, 3, ...}, and the resource model set of each service resource is J = {1, 2, 3, ...}, then the resource type set and resource model set of each requested service resource are integrated to generate a service resource set X = {x ij The service resource set consists of multiple service resources. This historical customer information, its corresponding service resource set, and the processing time of that historical resource set are then combined to form a historical order data entry.
[0064] In one embodiment, frequent items of various service resource categories are extracted in advance from each historical order data according to the Apriori algorithm. For example, if the historical order data involves service resource categories X1, X2, X3...Xn, the Apriori algorithm can extract one of the frequent items of a category that appears multiple times in each historical order data, i.e., the frequent item might be {X2, X3, X4}. Figure 2 As shown.
[0065] Then, extract the resource model combinations corresponding to the frequent category items {X2, X3, X4} from the historical order data. For example, in a certain historical order data, the resource model combinations corresponding to X2, X3, and X4 are 1, 2, and 3 respectively, then the resource model combination corresponding to the frequent category item {X2, X3, X4} is {1, 2, 3}. For instance, suppose X2 is a storage resource, X3 is a memory resource, and X3 is a CPU resource. If a certain historical order data contains storage resources, memory resources, and CPU resources simultaneously, and the storage resource recorded in this historical order data is 2G, the memory resource is 4G, and the CPU resource is 6G, then the resource model combination corresponding to the frequent category item {storage resource, memory resource, CPU resource} is {2G, 4G, 6G}.
[0066] From each historical order data entry, find the resource model combinations corresponding to frequent items of that type. The set of all resource model combinations found is the resource model combination set. For example... Figure 3 As shown in the figure, {X2, X3, X4} on the left side represent frequently occurring items, and each path on the right side represents the resource model combination corresponding to the frequently occurring item {X2, X3, X4} found in historical order data. That is, the resource model combinations corresponding to {X2, X3, X4} include {1, 1, 1}, {1, 2, 1}, {1, 2, 2}, {2, 1, 1}, etc.
[0067] After extracting the resource model combination sets corresponding to each frequent item, for any type of frequent item, such as {X2, X3, X4}, its corresponding resource model combination set is obtained as the target resource model combination set. Then, the frequency of any resource model combination in the target resource model combination set is detected among all resource model combination sets. For example, the frequency of the resource model combination {3, 3, 2} in all resource model combination sets in the target resource model combination set corresponding to {X2, X3, X4} is detected. If its frequency is greater than a preset value, it is considered a frequent resource model item. Figure 3 The frequent items in the column are combined with the frequent items of category {X2, X3, X4} to form the frequent items of resource LM = {X23, X33, X42}, as shown below. Figure 4 As shown. For example, assuming {X2, X3, X4} represents {storage resources, memory resources, CPU resources}, and {3, 3, 2} represents {2G, 4G, 6G}, then the frequently occurring resource items can be determined as {storage resources - 2G, memory resources - 4G, CPU resources - 6G}.
[0068] By pre-extracting frequent items by category and then extracting corresponding frequent items by model based on the frequent items by category, frequent items by resource can be determined. This eliminates the need to traverse all service resources in historical order records to search for frequent items by resource, effectively reducing the search space and thus improving the efficiency of searching for frequent items by resource.
[0069] After identifying the frequent resource items, a first deep learning model, Model1, is established based on the set of frequent resource items L = {L1, L2, ..., LM} and the historical customer information G = {g1, g2, ..., gm}. Here, g1, g2, ..., gm are elements from the historical customer information, including the application area and type of the requested cloud service. Then, the historical customer information G is iteratively input into the first deep learning model, Model1, to output the corresponding frequent resource item Ln and adjust its parameters until the similarity between the output frequent resource item Ln and the service resources requested by the historical customer information meets a preset requirement. If the similarity exceeds the set value, the first deep learning model, Model1, is considered to have completed training.
[0070] In one embodiment, after obtaining the set of frequent resource items L, a second deep learning model Model2 is constructed based on the set of frequent resource items L, historical customer information G, and the processing time of each frequent resource item recorded in the historical order data. Then, the historical customer information G and its corresponding frequent resource item Ln are iteratively combined and input into the second deep learning model Model2, so that the second deep learning model Model2 outputs the corresponding processing time and adjusts the parameters until the time difference between the output processing time and the actual processing time in the historical order data that records the historical customer information G and its corresponding frequent resource item Ln meets a preset requirement. If the time difference is greater than the set time difference, it is determined that the second deep learning model Model2 has completed training.
[0071] In one embodiment, after training the first and second deep learning models, the current user information for the requested service resource is input into the trained first deep learning model to obtain at least one frequent target resource item matching the current user information from the frequent item set. Then, each frequent target resource item is input into the trained second deep learning model to determine the service processing time for that item. This allows the model to predict the service processing time of the service resources the user might request, providing the user with a psychological expectation of the processing time. Furthermore, the frequent target resource item with the shortest processing time is extracted from the frequent target resource items for service recommendation, thereby reducing application time.
[0072] To make the recommended target resources more accurate, in one embodiment, the frequency of the resource frequently occurring in the historical orders is greater than a preset frequency.
[0073] In one embodiment, after determining the frequent resource items based on the frequent model items and the frequent category items corresponding to the target resource model combination set, the frequency of the frequent resource item in each historical order data of the historical orders is detected; if its frequency of occurrence is greater than the preset frequency, the frequent resource item is retained; otherwise, the frequent resource item is deleted.
[0074] Considering that when applying for cloud service resources, it is common to apply for two or more resources simultaneously, such as when applying for visualization analysis data services, storage, memory, and CPU resources must be selected at the same time. Therefore, to avoid the final recommended target resource being rejected, in one embodiment, the frequently requested resource includes at least one set of coupled resources.
[0075] The coupled resources are the service resources that must be applied for simultaneously when applying for cloud service resource onboarding.
[0076] For example, such as Figure 2 If X3 and X4 are a set of coupled resources, then if each frequent resource item contains either X3 or X4, then it must also contain either X4 or X3. Furthermore, each frequent resource item must satisfy resource coupling.
[0077] Considering that only one model of each cloud service resource can be selected when submitting an application, in order to further avoid the final recommended target resource being rejected, in one embodiment, the service resources in the frequent resource items are of different types. That is, the service resources in the frequent resource items must have model exclusivity.
[0078] In one embodiment, after determining the service acceptance time of each frequent item of the target resource, the method further includes:
[0079] The service processing time of each of the target resources with frequent occurrences is pushed to the user terminal corresponding to the current user information.
[0080] By pushing the service processing time of each frequently requested target resource to the user's terminal, users can know the service processing time of each frequently requested target resource, allowing them to have a psychological expectation of the service processing time for resource requests. When not selecting recommended frequently requested target resources, users can choose the frequently requested target resources with a service processing time that is acceptable to them based on the service processing time of each frequently requested target resource, without having to select multiple service resources individually, thus reducing the time spent on service requests due to selecting service resources and improving the user experience.
[0081] The cloud service recommendation device provided by the present invention is described below. The cloud service recommendation device described below can be referred to in correspondence with the cloud service recommendation method described above.
[0082] In one embodiment, such as Figure 5 As shown, a cloud service recommendation device is provided, including:
[0083] The frequent item determination module 210 is used to input the current user information into the trained first deep learning model, so as to determine at least one target resource frequent item that matches the current user information from the frequent item set formed by the service resources of historical orders according to the first deep learning model;
[0084] The duration determination module 220 is used to input each frequent item of the target resource into the trained second deep learning model to determine the service acceptance duration of each frequent item of the target resource.
[0085] Service recommendation module 230 is used to recommend services based on the service acceptance time of each of the target resource frequent items, by obtaining the target resource frequent item with the shortest service acceptance time.
[0086] The first deep learning model is trained using historical customer information from the historical orders and the set of frequent items. The set of frequent items includes multiple frequent resource items, and the frequent resource items include multiple service resources.
[0087] The second deep learning model is trained using the historical customer information, the set of frequent items, and the service processing time of each service resource in the historical orders.
[0088] In one embodiment, the frequent item determination module 210 is further configured to:
[0089] Based on the Apriori algorithm, frequent items are extracted from the resource types of each service resource in historical orders to obtain frequent items of each type;
[0090] Based on the frequent items of each category, extract the set of resource model combinations corresponding to each frequent item of each category from the historical orders;
[0091] Based on the frequency of occurrence of any resource model combination in the target resource model combination set in each resource model combination set, extract the frequent resource model items whose frequency of occurrence is greater than a preset threshold from the target resource model combination set.
[0092] The frequent items of resources are determined based on the frequent items of the model and the frequent items of the type corresponding to the combination set of the target resource models;
[0093] The first deep learning model is built and trained based on the set of frequent resource items and historical customer information.
[0094] The resource model combination includes multiple resource models.
[0095] In one embodiment, the resource type includes at least one of storage, memory, and CPU.
[0096] In one embodiment, the service recommendation module 230 is further configured to:
[0097] The service processing time of each of the target resources with frequent occurrences is pushed to the user terminal corresponding to the current user information.
[0098] In one embodiment, the frequency of occurrence of the resource frequent item in the historical orders is greater than a preset frequency.
[0099] In one embodiment, the frequent resource items include at least one set of coupled resources;
[0100] The coupled resources are the service resources that must be applied for simultaneously when applying for cloud service resource onboarding.
[0101] In one embodiment, the service resources in the frequent resource items are of different types.
[0102] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other via the communication bus 840. The processor 810 can call a computer program in the memory 830 to execute the steps of the cloud service recommendation method, such as including:
[0103] The current user information is input into the trained first deep learning model, so that, based on the first deep learning model, at least one target resource frequent item that matches the current user information is determined from the frequent item set formed by the service resources of historical orders.
[0104] Input each frequent item of the target resource into the trained second deep learning model to determine the service acceptance time of each frequent item of the target resource.
[0105] Based on the service acceptance time of each of the target resource frequent items, the target resource frequent item with the shortest service acceptance time is selected for service recommendation;
[0106] The first deep learning model is trained using historical customer information from the historical orders and the set of frequent items. The set of frequent items includes multiple frequent resource items, and the frequent resource items include multiple service resources.
[0107] The second deep learning model is trained using the historical customer information, the set of frequent items, and the service processing time of each service resource in the historical orders.
[0108] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0109] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the steps of the cloud service recommendation method provided in the above embodiments, such as including:
[0110] The current user information is input into the trained first deep learning model, so that, based on the first deep learning model, at least one target resource frequent item that matches the current user information is determined from the frequent item set formed by the service resources of historical orders.
[0111] Input each frequent item of the target resource into the trained second deep learning model to determine the service acceptance time of each frequent item of the target resource.
[0112] Based on the service acceptance time of each of the target resource frequent items, the target resource frequent item with the shortest service acceptance time is selected for service recommendation;
[0113] The first deep learning model is trained using historical customer information from the historical orders and the set of frequent items. The set of frequent items includes multiple frequent resource items, and the frequent resource items include multiple service resources.
[0114] The second deep learning model is trained using the historical customer information, the set of frequent items, and the service processing time of each service resource in the historical orders.
[0115] On the other hand, embodiments of this application also provide a processor-readable storage medium storing a computer program for causing a processor to perform the steps of the methods provided in the above embodiments, such as including:
[0116] The current user information is input into the trained first deep learning model, so that, based on the first deep learning model, at least one target resource frequent item that matches the current user information is determined from the frequent item set formed by the service resources of historical orders.
[0117] Input each frequent item of the target resource into the trained second deep learning model to determine the service acceptance time of each frequent item of the target resource.
[0118] Based on the service acceptance time of each of the target resource frequent items, the target resource frequent item with the shortest service acceptance time is selected for service recommendation;
[0119] The first deep learning model is trained using historical customer information from the historical orders and the set of frequent items. The set of frequent items includes multiple frequent resource items, and the frequent resource items include multiple service resources.
[0120] The second deep learning model is trained using the historical customer information, the set of frequent items, and the service processing time of each service resource in the historical orders.
[0121] The processor-readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to magnetic memory (e.g., floppy disk, hard disk, magnetic tape, magneto-optical disk (MO)), optical memory (e.g., CD, DVD, BD, HVD), and semiconductor memory (e.g., ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)).
[0122] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0123] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.
Claims
1. A cloud service recommendation method, characterized by, include: The current user information is input into the trained first deep learning model, so that, based on the first deep learning model, at least one target resource frequent item that matches the current user information is determined from the frequent item set formed by the service resources of historical orders. Input each frequent item of the target resource into the trained second deep learning model to determine the service acceptance time of each frequent item of the target resource. Based on the service acceptance time of each of the target resource frequent items, the target resource frequent item with the shortest service acceptance time is selected for service recommendation; The first deep learning model is trained using historical customer information from the historical orders and the set of frequent items. The set of frequent items includes multiple frequent resource items, and the frequent resource items include multiple service resources. The second deep learning model is trained using the historical customer information, the set of frequent items, and the service processing time of each service resource in the historical orders. The method further includes: Based on the Apriori algorithm, frequent items are extracted from the resource types of each service resource in historical orders to obtain frequent items of each type; Based on the frequent items of each category, extract the set of resource model combinations corresponding to each frequent item of each category from the historical orders; Based on the frequency of occurrence of any resource model combination in the target resource model combination set in each resource model combination set, extract the frequent resource model items whose frequency of occurrence is greater than a preset threshold from the target resource model combination set. The frequent items of resources are determined based on the frequent items of the model and the frequent items of the type corresponding to the combination set of the target resource models; The first deep learning model is built and trained based on the set of frequent resource items and historical customer information. The resource model combination includes multiple resource models. 2.The cloud service recommendation method of claim 1, wherein, The resource types include at least one of storage, memory, and CPU. 3.The cloud service recommendation method of claim 1, wherein, Also includes: The service processing time of each of the target resources with frequent occurrences is pushed to the user terminal corresponding to the current user information. 4.The cloud service recommendation method according to any one of claims 1-3, characterized in that, The frequency of the resource frequent item in the historical orders is greater than the preset frequency. 5.The cloud service recommendation method according to any one of claims 1-3, wherein, The frequent resource items include at least one set of coupled resources; The coupled resources are the service resources that must be applied for simultaneously when applying for cloud service resource onboarding.
6. The cloud service recommendation method according to any one of claims 1-3, characterized in that, The service resources in the frequent resource items are of different types.
7. A cloud service recommendation device, comprising: The frequent item determination module is used to input the current user information into a trained first deep learning model, so as to determine at least one target resource frequent item that matches the current user information from the frequent item set formed by the service resources of historical orders according to the first deep learning model; The duration determination module is used to input the frequent items of each target resource into the trained second deep learning model to determine the service acceptance duration of each frequent item of the target resource. The service recommendation module is used to recommend services based on the service acceptance time of each of the target resource frequent items, and to obtain the target resource frequent item with the shortest service acceptance time. The first deep learning model is trained using historical customer information from the historical orders and the set of frequent items. The set of frequent items includes multiple frequent resource items, and the frequent resource items include multiple service resources. The second deep learning model is trained using the historical customer information, the set of frequent items, and the service processing time of each service resource in the historical orders. The cloud service recommendation device is also used for: Based on the Apriori algorithm, frequent items are extracted from the resource types of each service resource in historical orders to obtain frequent items of each type; Based on the frequent items of each category, extract the set of resource model combinations corresponding to each frequent item of each category from the historical orders; Based on the frequency of occurrence of any resource model combination in the target resource model combination set in each resource model combination set, extract the frequent resource model items whose frequency of occurrence is greater than a preset threshold from the target resource model combination set. The frequent items of resources are determined based on the frequent items of the model and the frequent items of the type corresponding to the combination set of the target resource models; The first deep learning model is built and trained based on the set of frequent resource items and historical customer information. The resource model combination includes multiple resource models.
8. An electronic device comprising a processor and a memory having a computer program stored therein, characterized in that, When the processor executes the computer program, it implements the steps of the cloud service recommendation method according to any one of claims 1 to 6.
9. A computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the cloud service recommendation method according to any one of claims 1 to 6.