Method, device, readable storage medium and program product for performing network service

Abstract

Embodiments of the present application provide a kind of algorithm network service execution method, equipment, readable storage medium and program product, it is related to computing power network technical field.The method comprises: receiving the service demand information of user for target algorithm network service, and based on service demand information extraction contains user experience quality service demand characteristic information as user intent;Determine the service intent corresponding to user intent, service intent includes the service level target value of multiple service level indicators corresponding to user intent, and based on the service level target value of multiple service level indicators, determine target algorithm network service service strategy, to make algorithm network service system based on target algorithm network service service strategy executes target algorithm network service, so specifically determine target algorithm network service service strategy, fully consider the service experience of user, to effectively improve the service quality in the process of algorithm network service execution.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computing power network, and particularly relates to a computing power network service execution method and device, a readable storage medium and a program product. BACKGROUND

[0002] Computing power network resources mainly include distributed computing power network resources and connected computing power network resources. The computing power network resources can be efficiently interconnected through network infrastructure based on all-optical chassis and Internet Protocol Address (IP) bearing technology, so as to meet the needs of computing power network service execution.

[0003] In order to ensure the quality of service of the computing power network service, the traditional quality of service guarantee is from the perspective of commercial agreement, mainly for the function and performance management of the service level agreement (SLA) signed with the customer, and the target value of the SLA is set in combination with the service level objective (SLO) to guarantee the quality of service. Among them, the SLA mainly includes service function provision, service scale, service quality, service measurement, guarantee level and penalty, etc.

[0004] However, the above-mentioned quality of service guarantee is mainly realized through function and performance management, and the service experience (QoE) of the user is not fully considered, so as to cause the deviation of the quality of service in the computing power network service execution process. SUMMARY

[0005] The present application provides a computing power network service execution method, device, readable storage medium and program product, which fully considers the service experience of the user, so as to effectively improve the quality of service in the computing power network service execution process.

[0006] In a first aspect, the present application provides a computing power network service execution method, comprising:

[0007] receiving service demand information of a user for a target computing power network service, and extracting service demand feature information containing user experience quality as user intention based on the service demand information;

[0008] determining a service intention corresponding to the user intention, the service intention including service level objective values of a plurality of service level indicators corresponding to the user intention;

[0009] determining a target computing power network service strategy based on the service level objective values of the plurality of service level indicators; wherein the target computing power network service strategy is used for the computing power network service system to execute the target computing power network service.

[0010] According to the network service execution method provided in the application, the service demand characteristic information containing the user experience quality is extracted based on the service demand information as the user intention, including:

[0011] The service demand information is input into a machine learning model, and the service demand information is inferred and output to obtain an inference result;

[0012] The service demand characteristic information containing the user experience quality in the inference result is determined as the user intention.

[0013] According to the network service execution method provided in the application, the service demand characteristic information containing the user experience quality is extracted based on the service demand information as the user intention, including:

[0014] The service demand information is converted into a network configuration language or a programming language;

[0015] Based on the network configuration language or the programming language, the service demand characteristic information containing the user experience quality is extracted as the user intention.

[0016] According to the network service execution method provided in the application, the service intention corresponding to the user intention is determined, including:

[0017] Based on a pre-constructed matching model between the user intention and the service quality target, the service intention corresponding to the user intention is determined.

[0018] According to the network service execution method provided in the application, the result of the network service system executing the target network service includes service level actual values of the plurality of service level indicators, and the method further includes:

[0019] Determine whether the difference between the service level actual value corresponding to each service level indicator and the service level target value corresponding to each service level indicator is within a preset range;

[0020] In response to any of the difference values not being within the preset range, a service adjustment strategy corresponding to the service level indicator is determined; the network service system is controlled to execute the target network service based on the service adjustment strategy until the difference between the service level actual value and the service level target value is within the preset range.

[0021] According to the network service execution method provided in the application, the method further includes:

[0022] In response to the difference value not being within the preset range, alarm information corresponding to the service level indicator exceeding the preset range of difference value is generated.

[0023] According to the network service calculation method provided in the application, the target network service calculation service strategy is determined based on the service level target values of the plurality of service level indicators, and the method comprises the following steps:

[0024] At least one network service calculation service strategy is determined based on the service level target values of the plurality of service level indicators.

[0025] The service level historical values of the plurality of service level indicators are obtained.

[0026] The service level historical values of the plurality of service level indicators and the at least one network service calculation service strategy are input into a machine learning prediction algorithm model to obtain service level prediction values of the plurality of service level indicators and network resource state indicators corresponding to each network service calculation service strategy.

[0027] The target network service calculation service strategy is determined from the at least one network service calculation service strategy based on the service level target values of the plurality of service level indicators, the service level prediction values and the network resource state indicators corresponding to each network service calculation service strategy.

[0028] According to the network service calculation method provided in the application, the target network service calculation service strategy is determined based on the service level target values of the plurality of service level indicators, and the method comprises the following steps:

[0029] The system function indicators and system performance indicators of the network service calculation system are obtained.

[0030] The network service calculation service strategy is determined based on the service level target values of the plurality of service level indicators, the system function indicators and the system performance indicators.

[0031] According to the network service calculation method provided in the application, after the target network service calculation is performed based on the target network service calculation service strategy, the method further comprises the following steps:

[0032] The service level update values of the plurality of service level indicators and the network resource update states of the network service calculation system are obtained.

[0033] The evaluation value of the target network service calculation service strategy is determined based on the service level update values of the plurality of service level indicators and the network resource update states of the network service calculation system.

[0034] In a second aspect, the application provides a network service calculation device, which comprises:

[0035] A receiving unit is configured to receive service demand information of a target network service calculation from a user.

[0036] extracting, by an extraction unit, service demand feature information containing user experience quality as user intention based on the service demand information;

[0037] determining, by a first processing unit, a service intention corresponding to the user intention, the service intention including service level target values of multiple service level indicators corresponding to the user intention;

[0038] determining, by a second processing unit, a target networked business service policy based on the service level target values of the multiple service level indicators, wherein the target networked business service policy is used for a networked business system to execute the target networked business.

[0039] According to the networked business execution device provided in the present application, the extraction unit is configured to extract service demand feature information containing user experience quality as user intention based on the service demand information, and the extraction unit includes:

[0040] inputting the service demand information into a machine learning model, performing inference on the service demand information, and obtaining an inference result;

[0041] determining the service demand feature information containing user experience quality in the inference result as the user intention.

[0042] According to the networked business execution device provided in the present application, the extraction unit is configured to extract service demand feature information containing user experience quality as user intention based on the service demand information, and the extraction unit includes:

[0043] translating the service demand information into a network configuration language or a programming language;

[0044] extracting the service demand feature information containing user experience quality as the user intention based on the network configuration language or the programming language.

[0045] According to the networked business execution device provided in the present application, the first processing unit is configured to determine a service intention corresponding to the user intention, and the first processing unit includes:

[0046] determining the service intention corresponding to the user intention based on a pre-constructed matching model between user intention and service quality target.

[0047] According to the networked business execution device provided in the present application, a result of the networked business system executing the target networked business includes service level actual values of the multiple service level indicators, and the device further includes:

[0048] determining, by a third processing unit, whether a difference between a service level actual value corresponding to each service level indicator and a service level target value corresponding to each service level indicator is within a preset range.

[0049] a fourth processing unit, configured to determine a service adjustment strategy corresponding to the service level indicator in response to any of the difference values not being in the preset range;

[0050] a control unit, configured to control the network service system to execute the target network service based on the service adjustment strategy until the difference value between the actual value of the service level and the target value of the service level is in the preset range.

[0051] According to the network service execution device provided in the present application, the device further comprises:

[0052] a generating unit, configured to generate alarm information of the service level indicator exceeding the preset range of the difference value in response to the difference value not being in the preset range.

[0053] According to the network service execution device provided in the present application, the second processing unit is configured to determine a target network service strategy based on the target values of the plurality of service level indicators, comprising:

[0054] determining at least one network service strategy based on the target values of the plurality of service level indicators;

[0055] obtaining historical values of the plurality of service level indicators;

[0056] inputting the historical values of the plurality of service level indicators and the at least one network service strategy into a machine learning prediction algorithm model to obtain predicted values of the plurality of service level indicators and network resource state indicators corresponding to each network service strategy;

[0057] determining the target network service strategy from the at least one network service strategy based on the target values of the plurality of service level indicators, the predicted values of the plurality of service level indicators and the network resource state indicators corresponding to each network service strategy.

[0058] According to the network service execution device provided in the present application, the second processing unit is configured to determine a plurality of network service strategies based on the target values of the plurality of service level indicators, comprising:

[0059] obtaining system function indicators and system performance indicators of the network service system;

[0060] determining the network service strategies based on the target values of the plurality of service level indicators, the system function indicators and the system performance indicators.

[0061] According to the network calculation service execution device provided in the application, after the target network calculation service is executed based on the target network calculation service service policy, the device further comprises:

[0062] The acquisition unit is configured to acquire service level update values of the multiple service level indexes and network resource update states of the network calculation service system.

[0063] The fifth processing unit is configured to determine an evaluation value of the target network calculation service service policy based on the service level update values of the multiple service level indexes and the network resource update states of the network calculation service system.

[0064] In a third aspect, the embodiments of the application further provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the network calculation service execution method according to any one of the first aspect when executing the program.

[0065] In a fourth aspect, the embodiments of the application further provide a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executable on a processor to implement the network calculation service execution method according to any one of the first aspect.

[0066] In a fifth aspect, the embodiments of the application further provide a computer program product, which includes a computer program, and the computer program is executable on a processor to implement the network calculation service execution method according to any one of the first aspect.

[0067] The network calculation service execution method, device, readable storage medium and program product provided in the embodiments of the application receive service demand information of a user for a target network calculation service, extract service demand characteristic information including user experience quality QoE as user intention based on the service demand information, determine a service intention corresponding to the user intention, the service intention includes service level target values of multiple service level indexes corresponding to the user intention, and determine a target network calculation service service policy based on the service level target values of the multiple service level indexes, so that the network calculation service system executes the target network calculation service based on the target network calculation service service policy. In this way, the service demand characteristic information including user experience quality QoE is extracted as user intention based on the service demand information, the service intention meeting service quality demand is constructed based on the user intention, the service intention includes service level target values of multiple service level indexes, the target network calculation service service policy is determined in a targeted manner, the service experience of the user is fully considered, and therefore the service quality in the network calculation service execution process is effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0068] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and for those skilled in the art, other drawings can be obtained from these drawings without creative labor.

[0069] Figure 1 A framework schematic diagram of an algorithm network service system is provided for the embodiments of the application.

[0070] Figure 2 A flow schematic diagram of an algorithm network service execution method is provided for the embodiments of the application.

[0071] Figure 3 A framework schematic diagram of service quality assurance is provided for the embodiments of the application.

[0072] Figure 4 A flow schematic diagram of an intention-driven service quality assurance closed loop is provided for the embodiments of the application based on a plurality of service level indicators.

[0073] Figure 5 A structural schematic diagram of an algorithm network service execution device is provided for the embodiments of the application.

[0074] Figure 6 An entity structural schematic diagram of an electronic device is provided for the embodiments of the application. DETAILED DESCRIPTION

[0075] In order to make the purpose, technical solutions and advantages of the application clearer, the following will combine the drawings in the application to clearly and completely describe the technical solutions in the application. Obviously, the described embodiments are some embodiments of the application, not all embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0076] In the embodiments of the application, "at least one" means one or more, and "multiple" means two or more. The association relationship of the associated objects is described by "and / or", which means that there can be three relationships, for example, A and / or B, which means that there are three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In the textual description of the application, the character " / " generally represents an "or" relationship between the associated objects before and after it.

[0077] The technical scheme provided by the embodiments of the present application can be applied to an algorithm network service execution scene. In the algorithm network, a high-quality algorithm network service quality efficient guarantee driven by an intention is provided for a user, for example, east-to-west calculation, cloud-edge collaboration, intelligent video, smart factory, and the like. It can be understood that the technical scheme provided by the embodiments of the present application can also be extended to other complex end-to-end cross-domain algorithm network service system scenes, and an intention-driven service quality guarantee technology is introduced.

[0078] In order to guarantee the service quality of the algorithm network service, the traditional service quality guarantee is mainly from the perspective of a business agreement, mainly performs function and performance management on the SLA signed with the customer, and sets the target value of the SLA implementation in combination with the SLO, so as to guarantee the service quality.

[0079] Among them, the SLA mainly includes service function provision, service scale, service quality, service measurement, guarantee level and penalty, and the like.

[0080] However, the above service quality guarantee is mainly realized by function and performance management, and the service experience of the user is not considered. The quality of service experience (QoE) is a comprehensive measurement of the subjective satisfaction of the user when using a specific service, for example, video streaming media, VoIP call, web browsing, and the like, which covers comprehensive evaluation of multiple dimensions such as content availability, service quality, interface friendliness and interactive response speed.

[0081] Compared with the existing function and performance management, the QoE pays more attention to the actual feeling and expectation of the user. Therefore, in order to improve the service quality in the algorithm network service execution process, the embodiments of the present application provide an algorithm network service execution method. In the algorithm network service execution process, the quality of service experience (QoE) is improved by fully considering the intention of the user, so as to improve the service quality in the algorithm network service execution process.

[0082] For example, refer to Figure 1 as shown, Figure 1 A framework schematic diagram of an algorithm network service system provided by the embodiments of the present application is provided. The algorithm network service system can include an algorithm network operation layer, an algorithm network management and control layer and an algorithm network resource layer. The algorithm network operation layer and the algorithm network management and control layer can be integrated in an algorithm server, and the algorithm network resource layer can be integrated in a networking device in the algorithm network.

[0083] In some cases, the computing network operation layer and the computing network management layer belong to the computing network arrangement management layer in the computing network architecture, and the computing power network resource layer belongs to the computing network infrastructure layer. The embodiments of the present application do not make special limitations here. In some cases, the computing network architecture includes a computing power service layer and a computing power routing layer. The computing power routing layer is connected to the computing power service layer, the computing network arrangement management layer, and the computing network infrastructure layer through an interface. The computing network arrangement management layer is also connected to the computing power service layer and the computing network arrangement management layer through an interface. The computing power service layer provides an interface for users to interact with the computing network business execution system, allowing users to input their requirements and expectations for businesses such as augmented reality (AR), virtual reality (VR), etc. in various forms, such as natural language, graphical interface, or domain-specific language, etc. These intentions can be understood, recognized, and represented by the computing network business system, which is the basis for subsequent service quality control. The computing power service layer receives user input business demand information for target computing network businesses in various ways that support multiple domains and scenarios, such as AR, VR, V2X (Vehicle to Everything), artificial intelligence (AI), etc. that facilitate user input and modification. In some cases, users can use specialized expression methods or programming instructions to perform intent representation operations based on the business characteristics and requirements of different domains. The embodiments of the present application do not make special limitations here.

[0084] In combination Figure 1 As shown in the computing network business system, the user intent management module in the computing network operation layer can receive user input business demand information for a target computing network business, and extract business demand feature information containing quality of experience (QoE) as user intent based on the input business demand information. Determine the service intent corresponding to the user intent, which includes the service level target value of the multiple service level indicators corresponding to the user intent, and determine the target computing network business service strategy based on the service level target value of the multiple service level indicators, so that the computing network business system executes the target computing network business based on the target computing network business service strategy. In this way, the business demand feature information containing the quality of experience of the user is extracted as the user intent based on the user input business demand information, and the service intent that meets the service quality requirements and includes the service level target value of the multiple service level indicators is constructed based on the user intent. The target computing network business service strategy is determined in a targeted manner, fully considering the service experience of the user, thereby effectively improving the service quality in the computing network business execution process.

[0085] In the following, the computing network business execution method provided by the present application will be described in detail through the following specific embodiments. It can be understood that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may

[0086] Figure 2 A flowchart of an algorithmic network service execution method provided in an embodiment of the present application is shown. The method can be executed by software and / or hardware devices. For example, as shown in the flowchart, the algorithmic network service execution method can include the following steps. Figure 1

[0087] S201, receiving service demand information of a user for a target algorithmic network service, and extracting service demand characteristic information containing user experience quality as user intention based on the service demand information.

[0088] The user intention is an algorithmic network intention expectation issued by the user and / or service in the service quality guarantee of the algorithmic network, and is the demand and expectation of the user for the algorithmic network service. The user intention mainly focuses on the perceived demand of the algorithmic network service, the continuity of the service, and the comprehensive experience of the user for the service, and is extracted from the service demand information of the user for the target algorithmic network service. The service demand characteristic information containing user experience quality is the target and requirement of the algorithmic network defined by the algorithmic network service system from the algorithmic network service level, i.e., the service to be completed by the algorithmic network in the management domain of the algorithmic network provider and the specific requirements in the service. The user intention has various expression forms, such as natural language, voice, graphical user interface, domain-specific language, etc. Alternatively, the input of the user intention can be changed from natural language to voice, picture, or other forms, which can be set according to actual needs.

[0089] For example, in the embodiment of the present application, the experience quality QoE in the user intention mainly includes multiple service level indicators, which can include registration delay, one-way delay and round-trip time (RTT), synchronization loss time, damage duration, frame rate, cache level, etc., which can be set according to actual needs.

[0090] The registration delay represents the time consumed from the start of the application program to the entry into the service through loading, which can include the response delay of the server and the total two-way delay of network forwarding. Generally, a shorter registration delay means that the user can start enjoying the application service faster, improving the user retention rate. Correspondingly, the service level indicator SLI or the service quality QoS can include the response delay of the server and the total two-way delay of network forwarding. A shorter registration delay means that the user can start enjoying the application service faster, improving the user retention rate.

[0091] ​One-way delay and round-trip time include uplink one-way delay, downlink one-way delay, and delay in the RTT communication network. Different processes have different processing delay requirements; uplink delay requirements, downlink delay requirements, etc. of network forwarding directly affect the real-time interaction experience of users and services. If the delay is too long, it may cause problems such as interruption of voice or video calls, lag in game operation, etc., which seriously affect the user experience. Correspondingly, the service level indicator SLI or the quality of service QoS can include: different processing delay requirements for different processes; and uplink and downlink delay requirements of network forwarding. Among them, the service level indicator SLI directly affects the real-time interaction experience of users and services. If the delay is too long, it may cause problems such as interruption of voice or video calls, lag in game operation, etc., which seriously affect the user experience.

[0092] Synchronization loss time refers to the phenomenon of audio-visual asynchronization when the absolute difference between the playback time of the last playback frame of the video stream and the playback time of the last playback frame of the voice / audio stream is greater than the synchronization threshold. It is mainly reflected in the synchronization of multiple service streams, different service network delays and priority requirements, etc. In a video conference, if voice and picture do not match, it may cause communication obstacles. Correspondingly, the service level indicator SLI or the quality of service QoS can include: synchronization of multiple service streams and different service network delays and priority requirements. In a video conference, if voice and picture do not match, it may cause communication obstacles.

[0093] Damage duration refers to the NPT time interval from the last good frame to the first subsequent good frame. It is mainly reflected in the packet loss requirements for service streams, requirements for network packet loss, etc. The damaged frame can be a completely lost frame or a media frame with degraded quality. For streaming media applications, too long damage duration may cause problems such as video stuttering, audio distortion, etc. Correspondingly, the service level indicator SLI or the quality of service QoS can include: packet loss requirements for service streams and requirements for network packet loss. Among them, the damaged frame can be a completely lost frame or a media frame with degraded quality. For streaming media applications, too long damage duration may cause problems such as video stuttering, audio distortion, etc.

[0094] Frame rate, i.e. playback frame, refers to the number of frames displayed during the measurement period divided by the measurement period. It is mainly reflected in the requirements for service side cache size and strategy, network bandwidth, etc. If the frame rate is too low, it may cause problems such as video picture stuttering, unsmoothness, and reduced user viewing experience. Correspondingly, the service level indicator SLI or the quality of service QoS can include: requirements for service side cache size and strategy, and network bandwidth. If the frame rate is too low, it may cause problems such as video picture stuttering, unsmoothness, and reduced user viewing experience.

[0095] The cache level refers to the time during which the media data of all active media components is available for continuous playing starting from the current playing time. The cache level is mainly embodied in the service end cache size and policy requirements, network bandwidth requirements, and the like. A higher cache level means that the application can still maintain good playing experience in the case of unstable network and reduce playing interruption caused by network problems. Correspondingly, the service level indicator SLI or the quality of service QoS can include the service end cache size and policy requirements and the network bandwidth requirements. A higher cache level means that the application can still maintain good playing experience in the case of unstable network and reduce playing interruption caused by network problems.

[0096] For example, in some embodiments of the present application, when determining the user intention of the user based on the service demand information of the target algorithm network service, the user intention can be determined based on the service demand information of the target algorithm network service, as shown in Figure 3 Figure 3 A service quality guarantee framework provided by the embodiments of the present application is shown in the figure. The service demand information input by the user is received through the interactive interface, the service demand feature information containing the quality of experience QoE is extracted based on the service demand information to obtain the user intention, for example, the service demand information can be input into a machine learning model, the service demand information is inferred and output to obtain an inference result, and the service demand feature information containing the quality of experience QoE in the inference result is determined as the user intention.

[0097] In some embodiments of the present application, the service demand information input by the user can have various expression forms, such as natural language, voice, graphical user interface, domain-specific language, and the like. The acquisition of the user intention can be performed in real time to ensure that the user intention is applied to the algorithm network service system in time.

[0098] For example, the machine learning model can be a multi-modal model, and the multi-modal model can be, for example, as shown in the attached Figure 1 ​The natural language processing (NLP) model shown can be a support vector machine (SVM), a neural network classification model, a hidden Markov model (HMM), a Gaussian mixture model (GMM), or other machine learning model used for intent recognition, which is not limited in the present application. For various forms of business demand information input by the user, a multi-modal model can be used. The multi-modal model can be a multi-modal model customized for user intent, a general multi-modal model, or a large language model, which is not limited in the present application. The user intent is determined by the multi-modal model, which takes the user experience QoE related business demand information as the service quality sensory metric of the user input intent, represents the user experience index metric of the user's service expectation, forms the quality part of the user demand intent in the intent driven closed loop of service quality, and can be abstracted as the satisfaction and dissatisfaction of the user's intent for the network demand, so as to determine the service level target value of the multiple service level indicators corresponding to the user intent. For example, the user input business demand information related to user experience QoE for the target network service input is: "provide computing power and network resources for XX cloud game to run smoothly every night from 8 pm to 10 pm", and the business feature demand information including "every night from 8 pm to 10 pm", "smooth running", "XX cloud game", "computing power", "network", etc. is obtained by analyzing and extracting the above business demand information by the natural language model as the user intent. Based on the historical experience library, the multiple service level indicators corresponding to the above user intent can include registration delay, one-way delay and round-trip time, damage duration, synchronization loss time, frame rate, cache level, etc. as service intent.

[0099] In some embodiments of the present application, user intent acquisition generally includes acquisition from natural language and domain-specific language. For example, the user's intent is identified and extracted from natural language by BiLSTM-CRF model for named entity recognition, which is trained on a training set and usually requires a large amount of labeled data. The user's configuration intent is acquired from natural language by analyzing the syntax structure of the sentence through a syntax tree, which requires domain-specific syntax rules and knowledge of the network. A pre-trained large language model such as RoBERTa is used in combination with a Dialogue State Tracking (DST) model (such as DIET) to implement entity recognition and semantic classification tasks. Or use a large language model in combination with a pre-defined prompt (Prompt) to directly acquire the user's intent from natural language, which can realize the dialogue with the user to understand and extract the user's query intent.

[0100] It can be understood that the resource requirements of the computing network are different in different scenarios, and the above models are continuously trained using user feedback and system performance data to adapt to different scenarios. Since the resource requirements are dynamically changing, the user intent acquisition has self-adaptability through model training, which can automatically adjust the acquisition strategy according to the user demand changes. The domain-specific language (DSL) allows users to express the configuration intention of the computing network in the form of functions or programming languages; here, the DSL similar to the programming language requires users to define the computing network configuration strategy in a programming manner, and users need to input a set of DSL, which may require users to have certain programming ability.

[0101] For example, in some embodiments of the present application, when extracting the service demand feature information containing the quality of user experience as the user intent based on the service demand information, the service demand can also be translated to generate a network configuration language or a programming language; and then based on the network configuration language or the programming language, the service demand feature information containing the quality of user experience is extracted as the user intent. The translation of the service demand includes the service in which the natural language is used as the intent carrier, that is, when the user input service demand information is in the form of natural language, the intent is extracted and understood from the user's natural language input, and is translated into a specific network configuration language or a programming language to obtain computer understandable and processable content, so that the computing network service system can automatically perform the required operation. The translation of the service demand at least includes natural language processing, intent classification, slot filling, intermediate language, mapping and translation rules, and comprehensive system integration. The above steps can be selected according to actual needs.

[0102] Specifically, the natural language processing is to analyze and understand the user input service demand information through a natural language processing model (NLP), which includes steps such as word segmentation, part-of-speech tagging, named entity recognition, and syntax analysis to extract the key information of the user intent. When the user intent input supports multi-modal input processing, the natural language processing model can be a multi-modal model. The multi-modal model can be a natural language processing (NLP) model or a Gaussian mixture model (GMM), and the multi-modal model can be a multi-modal model customized for user intent or a general multi-modal model, which is not particularly limited in the present application.

[0103] Intent Classification is the process of inputting user-entered business requirement information into a machine learning or deep learning model and classifying the analyzed text into specific user intents. This includes training a user intent classifier to map user inputs to predefined intent categories. Slot Filling is the process of identifying and extracting key information such as parameters, conditions, or entities from user intents. This information is typically represented in key-value pairs or other similar formats for subsequent processing.

[0104] Intermediate Language is the process of introducing an intermediate language or data structure to map user intents to higher-level network configuration languages or programming languages, which helps reduce the overall complexity of the algorithm network business system and improve maintainability.

[0105] Mapping and Translation Rules are developed by developing mapping rules or translation algorithms to translate the representation of the intermediate language into the target network configuration language or programming language. These rules can be based on algorithm network domain knowledge, templates, or machine learning models.

[0106] Integrated System Integration is the process of integrating the generated network configuration or programming language into a specific integrated system, such as an algorithm network business system, to perform actual network configuration changes or programming operations. This requires integration with existing systems, Application Programming Interface (API) calls, or script execution.

[0107] In some embodiments of the present application, some user intents require multiple algorithm network region resources, multiple scenarios, or other special conditions to achieve the user demand target. Based on the number of algorithm network resources and business scenarios of the user-entered business requirement information, the user intent is split into multiple sub-intents, and the above intent translation steps can also be performed for multiple sub-intents.

[0108] In some embodiments of the present application, the algorithm high concurrency handles user intent, and the intent-driven algorithm network control method can obtain a large number of random intents in multiple forms, multiple dimensions, and multiple scenarios in a short time. To ensure the quality of algorithm network services, it is necessary to quickly process high-concurrency intent translation capabilities. Here, algorithm resource lightweight collection, algorithm resource lightweight transmission, algorithm resource lightweight storage, and fast query are used to achieve high-concurrency user intent processing.

[0109] The lightweight collection of computing resource includes topology-based lightweight collection. The computing devices in the computing network are topologically divided, and a sketch is deployed in each sub-topology to collect computing resource information. The computing resource collection process is offloaded to the network instead of being directly transmitted to the computing network perception platform, thereby avoiding server overload and reducing network bandwidth occupation. The lightweight collection of computing resource also includes service demand-based collection. Different types of computing information are obtained according to different service demands. The corresponding sketch is selected for deployment to avoid transmitting all information of all computing devices, thereby reducing storage resource and network bandwidth occupation. For example, the computing network operator wants to obtain GPU devices with computing capacity greater than a certain threshold and accurate GPU computing capacity information. The sketch is used to obtain the corresponding GPU devices and their computing information. In this way, the non-GPU devices in the network and the GPU devices with computing capacity lower than the threshold are not transmitted, thereby reducing the burden of the computing network perception platform and network bandwidth occupation.

[0110] The lightweight transmission of computing resource refers to the technology of transmitting the computing information collected by the computing clusters in the sub-topology or different regions of the network to the computing network service management platform. The lightweight transmission of computing resource includes a sketch-based lightweight transmission strategy of computing resource. The computing information collected by the sub-topology or computing cluster is adaptively compressed according to the bandwidth and then transmitted to the computing network service manager. The adaptive network bandwidth can be realized by similar value counter merging, key compression and other technologies, and the real-time transmission of computing resource information can be realized. The lightweight transmission of computing resource also includes lightweight storage and fast query of computing resource.

[0111] The lightweight storage and fast query of computing resource includes a sketch-based lightweight storage and fast query update scheme. The sketch naturally has the advantage of small storage space occupation. The collected sketch only needs to be directly stored or decoded and stored in the sketch of the overall network view. On the other hand, in the query update, the computing information of the corresponding device can be quickly obtained by using the hash function of the sketch.

[0112] In some embodiments of the present application, when concurrent user intentions or multiple sub-user intentions occur, the priority of different service demands of the user intention or sub-user intention is managed, and the related demands of important services are preferentially processed to ensure the service quality of important services. Here, the improved scheduling strategy of computing network service strategy is used to realize timely and effective processing of computing network services according to different demands and priorities in the user intention.

[0113] For example, the priority of the service requirement is defined on the service operation side. The operator can set the priority according to the urgency of the input service requirement. Assuming that there are three services, VR service 1, VR service 2 and AR service, and the urgency of the service requirement of VR service 1 is the highest, the urgency of the service requirement of AR service is the second, and the urgency of the service requirement of VR service 2 is the lowest, the priority of VR service 1 can be set to the highest, for example, 0; the priority of AR service can be set to the second highest, for example, 1; and the priority of VR service 2 can be set to the lowest, for example, 2, so that the service quality of VR service 1 with the priority of 0 is satisfied first.

[0114] In some embodiments of the present application, after determining the user intention, the information of the user intention is visualized, for example, in a graphical form, so that the algorithm network administrator can more intuitively understand various service requirements in the algorithm network, thereby performing more effective management and adjustment. It should be noted that in the determination of the user intention, different service requirements are analyzed to realize intention association, and the mutual influence and dependency relationship between the service requirements in the user intention are found, thereby providing a basis for algorithm network planning and adjustment.

[0115] In some embodiments of the present application, after receiving the service requirement information of the user, the service requirement characteristic information containing the quality of experience (QoE) of the user is extracted as the user intention based on the service requirement information, and the following S202 is performed.

[0116] S202, determining a service intention corresponding to the user intention, the service intention including service level target values of a plurality of service level indicators corresponding to the user intention.

[0117] For example, in some embodiments of the present application, when determining the service intention corresponding to the user intention, a matching model between the user intention and the service quality target can be constructed in advance, and the service intention corresponding to the user intention can be determined based on the matching model. The service intention can include service level objective (SLO) values of a plurality of service level indicators, so that at least one algorithm network service strategy can be determined based on the service level target values.

[0118] The service level target values of the plurality of service level indicators are also the service quality guarantee indicators. For the case of multiple sub-intentions, the service intention of each sub-intention is determined, and the corresponding at least one target algorithm network service strategy is determined.

[0119] To realize cross-domain end-to-end collaboration, for example, in some embodiments of the present application, a service intent corresponding to the user intent can be determined, which can include service level target values of multiple service level indicators, i.e., to build a cross-domain end-to-end collaborative service quality assurance system. For example, based on a deep neural network algorithm, such as a multilayer perceptron (MLP) algorithm, the user intent can be combined with the service intent to determine the service intent corresponding to the user intent, which can include service level target values of multiple service level indicators, form a matching relationship containing the user intent and the service level target, and perform dynamic service quality assurance. In this way, by determining the service intent corresponding to the user intent, which can include service level target values of multiple service level indicators, a unified user experience indicator measurement can be formed, the automatic intent translation from the user intent to the service intent target can be realized, the user intent can be more efficiently and accurately captured, the foundation for efficient service quality assurance is laid, and thus a cross-domain service quality assurance system is realized. Figure 3 As shown, the service intent corresponding to the user intent is determined, which can include service level target values of multiple service level indicators, a matching relationship containing the user intent and the service level target is formed, and dynamic service quality assurance is performed. In this way, by determining the service intent corresponding to the user intent, which can include service level target values of multiple service level indicators, a unified user experience indicator measurement can be formed, the automatic intent translation from the user intent to the service intent target can be realized, the user intent can be more efficiently and accurately captured, the foundation for efficient service quality assurance is laid, and thus a cross-domain service quality assurance system is realized.

[0120] After determining the service level target values of multiple service level indicators, based on the service level target values of multiple service level indicators, at least one algorithm network business service strategy can be determined, i.e., S203 is executed as follows:

[0121] S203, based on the service level target values of multiple service level indicators, determine the target algorithm network business service strategy; wherein the target algorithm network business service strategy is used for the algorithm network business system to execute the target algorithm network business.

[0122] It is worth noting that in some embodiments of the present application, when executing the algorithm network business, the user's business demand information for the target algorithm network business is first received, and the business demand feature information containing the user experience quality (QoE) is extracted as the user intent based on the business demand information; the service intent corresponding to the user intent is determined, which includes service level target values of multiple service level indicators corresponding to the user intent, and based on the service level target values of multiple service level indicators, the target algorithm network business service strategy is determined, so that the algorithm network business system executes the target algorithm network business based on the target algorithm network business service strategy. In this way, based on the input demand, the business demand feature information containing the user experience quality (QoE) is extracted as the user intent, and the service intent satisfying the service quality demand constructed based on the user intent includes service level target values of multiple service level indicators, the target algorithm network business service strategy is determined, the service experience of the user is fully considered, and thus the service quality in the algorithm network business execution process is effectively improved.

[0123] The target algorithm network service strategy includes an algorithm network arrangement and scheduling strategy, so that the algorithm network resources are reasonably allocated, scheduled, migrated, etc., and cross-domain index decomposition is performed to meet the performance requirements and optimization targets of the user on the algorithm network service.

[0124] Based on the above Figure 1 In the embodiment shown in the above S103, the algorithm network service system can obtain corresponding execution results after executing the target algorithm network service. The execution results include service level actual values (SLOs) of multiple service level indicators. After obtaining the service level actual values of the multiple service level indicators, the service level actual values obtained, i.e., the service quality guarantee (SLO) indicator set, can be used to evaluate the user intention satisfaction degree and fed back to the user, so that an intention-driven service quality guarantee closed loop can be realized. For details, see the following Figure 4 Embodiment.

[0125] Based on the above Figure 1 After executing the target algorithm network service based on the target algorithm network service strategy, service level update values of multiple service level indicators and network resource update states of the algorithm network service system can be obtained. Based on the service level update values of the multiple service level indicators and the network resource update states of the algorithm network service system, the evaluation value of the target algorithm network service strategy is determined. In this way, by evaluating the target algorithm network service strategy, a closed loop control process is completed, so that the user can decide whether to modify or add other intentions according to his own needs, and then enter the next closed loop control cycle. Therefore, the target algorithm network service strategy can be determined more targetedly to ensure that the algorithm network can dynamically meet the service intention requirements and improve the service quality in the algorithm network service execution process.

[0126] For example, in some embodiments of the present application, the service level update values of the multiple service level indicators and the network resource update states of the algorithm network service system are obtained through algorithm network perception. After the target algorithm network service strategy is executed, the algorithm network perception perceives and monitors the state of the computing power network resources and the satisfaction of the business requirements, and feeds back the results to the user to realize end-to-end collaborative scheduling and optimization from the user side to the resource side. Specifically, it includes: collecting and updating the resource state information of a large number of devices in the algorithm network system through a control protocol or a detection mechanism, such as computing power type, computing power state, computing power capability, resource surplus, etc.; measuring and processing all network resource running states and transmission quality information through in-band telemetry and other network perception technologies, such as network bandwidth utilization rate, packet loss rate, delay, etc.; identifying and obtaining the satisfaction of the business computing power and network through access control, packet header encapsulation, service identification, etc. at the granularity of single business / single user, such as computing power service type, computing power SLI, network SLI, etc.

[0127] Figure 4A service level actual value based on multiple service level indicators is provided for an embodiment of the present application, and a flowchart of an intent-driven service quality guarantee closed loop is implemented. The method can include the following steps:

[0128] S401, determining whether a difference between a service level actual value corresponding to each service level indicator and a service level target value corresponding to each service level indicator is in a preset range.

[0129] The value of the preset range can be set according to actual needs. In this regard, the value of the preset range is not specifically limited in the embodiments of the present application.

[0130] For example, in some embodiments, for each service level indicator, the difference between the service level actual value corresponding to the service level indicator and the service level target value corresponding to the service level indicator needs to be calculated. If the difference between the service level actual value of the service level indicator and the service level target value of the service level indicator is in the preset range, it indicates that the satisfaction degree of the user intent is good, and the current service quality can meet the user demand. On the contrary, if the difference between the service level actual value of the service level indicator and the service level target value of the service level indicator is not in the preset range, it indicates that the satisfaction degree of the user intent is poor, and the current service quality cannot meet the user demand. Then, the following S402 is executed:

[0131] S402, in response to any difference not being in the preset range, determining a service adjustment strategy corresponding to the service level indicator.

[0132] The service adjustment strategy is used to optimize the execution of the target algorithm network business, so that the algorithm network business system can respond to the user intent in time and quickly give the service adjustment strategy, thereby improving the user experience. In the algorithm network business system, by analyzing the difference between the feedback result of the algorithm network business strategy execution in the last closed loop service period and the user intent target, a service quality guarantee (SLO) indicator set is formed and a series of action strategies are developed. Finally, the action strategies are converted into computer understandable and executable content, so as to narrow the gap between the current algorithm network business system state and the user intent state.

[0133] It should be noted that in the embodiments of the present application, as long as at least one difference between the differences corresponding to the service level indicators is not in the preset range, the service adjustment strategy corresponding to the service level indicator is determined.

[0134] Generally, during the execution of the target algorithm network business, the algorithm network state and the user intent are monitored in real time, and the algorithm network resources are configured and optimized through an automatic mechanism, so that in the case where any difference is not in the preset range, the service adjustment strategy corresponding to the service level indicator is determined, thereby meeting the user's business demand.

[0135] S403, control the algorithm network business system to execute the target algorithm network business based on the service adjustment strategy until the difference between the actual value of the service level and the target value of the service level is in the preset range.

[0136] It can be seen that, in some embodiments of the application, by comparing the actual value of the service level of the service level index with the target value of the service level of the service level index, and in the case that the difference between the actual value of the service level of the service level index and the target value of the service level of the service level index is not in the preset range, the control algorithm network business system executes the target algorithm network business based on the service adjustment strategy, which can realize the intent-driven service quality guarantee closed loop, realize the cross-domain end-to-end closed loop control from user intent to resource operation, and thus can effectively improve the subjective satisfaction of users.

[0137] For example, in some embodiments of the application, in response to the difference not being in the preset range, it indicates that the satisfaction degree of the user intent deviates, and the current service quality cannot meet the user demand, and alarm information corresponding to the service level index exceeding the preset range of the difference is generated to realize the alarm of the service level index exceeding the preset range of the difference.

[0138] Based on any of the above embodiments, based on the target values of the plurality of service level indexes, when determining the algorithm network business service strategy, in order to more accurately determine the algorithm network business service strategy, for example, in the embodiments of the application, at least one algorithm network business service strategy can be determined based on the target values of the plurality of service level indexes; the historical values of the plurality of service level indexes are obtained; the historical values of the plurality of service level indexes and at least one algorithm network business service strategy are input into a machine learning prediction algorithm model to obtain the predicted values of the plurality of service level indexes and the network resource state indexes corresponding to each algorithm network business service strategy; based on the target values of the plurality of service level indexes, the predicted values of the plurality of service level indexes, and the network resource state indexes corresponding to each algorithm network business service strategy, a target algorithm network business service strategy is determined from at least one algorithm network business service strategy, so that the network business service strategy is determined in combination with the predicted values of the plurality of service level indexes and the network resource state indexes corresponding to each algorithm network business service strategy, which can more accurately and specifically determine the algorithm network business service strategy, and lay a foundation for subsequent improvement of the service quality in the algorithm network business execution process.

[0139] For example, in some embodiments of the application, the network resource state index can include a network resource state index such as delay, bandwidth, jitter, etc., which can be set according to actual needs, and the embodiments of the application are not limited further herein.

[0140] For example, in some embodiments of the present application, the machine learning prediction algorithm model can be a regression prediction model, and can be a deep neural network prediction algorithm model, wherein the deep neural network prediction algorithm model can be a Long Short-Term Memory (LSTM) prediction algorithm model as shown in FIG. 8, and of course, can also be other deep neural network prediction algorithm models, such as a Convolutional Neural Network (CNN) prediction algorithm model, etc., which can be set according to actual needs. Figure 1

[0141] For example, in some embodiments of the present application, when determining the target algorithm network service strategy from at least one algorithm network service strategy based on the service level target values of the multiple service level indicators, the service level prediction values, and the network resource state indicators corresponding to each algorithm network service strategy, the service level target values of the multiple service level indicators, the service level prediction values, and the network resource state indicators corresponding to each algorithm network service strategy can be weighted, and the algorithm network service strategy corresponding to the maximum value after weighting is determined as the target algorithm network service strategy, thereby determining the target algorithm network service strategy.

[0142] Based on any of the above embodiments, based on the service level target values of the multiple service level indicators, the system function indicators and the system performance indicators of the algorithm network service system can also be obtained based on the service level target values of the multiple service level indicators, and based on the service level target values of the multiple service level indicators, the system function indicators and the system performance indicators, the algorithm network service strategy is determined together, which combines the user intent, the system function indicators and the system performance indicators to determine the algorithm network service strategy together, and can more accurately determine the algorithm network service strategy, thereby laying a foundation for improving the service quality in the subsequent algorithm network service execution process.

[0143] In the following, the algorithm network service execution device provided by the present application will be described, and the algorithm network service execution device described below can be mutually corresponding to the algorithm network service execution method described above.

[0144] Figure 5 For example, please refer to FIG. 9, which shows a structural schematic diagram of an algorithm network service execution device provided by an embodiment of the present application, which can include: Figure 5

[0145] The receiving unit 501 is configured to receive the service demand information of the target algorithm network service of the user.

[0146] ​​The extraction unit 502 is configured to extract service demand characteristic information containing user experience quality as user intention based on the service demand information.

[0147] The first processing unit 503 is configured to determine a service intention corresponding to the user intention, the service intention including service level target values of a plurality of service level indicators corresponding to the user intention.

[0148] The second processing unit 504 is configured to determine a target networked business service strategy based on the service level target values of the plurality of service level indicators, the target networked business service strategy being used for the networked business system to execute the target networked business.

[0149] For example, in some embodiments of the present application, the extraction unit 502 is configured to extract service demand characteristic information containing user experience quality as user intention based on the service demand information, including:

[0150] inputting the service demand information into a machine learning model, performing inference on the service demand information to obtain an inference result;

[0151] determining service demand characteristic information containing user experience quality in the inference result as the user intention.

[0152] For example, in some embodiments of the present application, the extraction unit 502 is configured to extract service demand characteristic information containing user experience quality as user intention based on the service demand information, including:

[0153] translating the service demand information into a network configuration language or a programming language;

[0154] extracting service demand characteristic information containing user experience quality as the user intention based on the network configuration language or the programming language.

[0155] For example, in some embodiments of the present application, the first processing unit 503 is configured to determine a service intention corresponding to the user intention, including:

[0156] determining the service intention corresponding to the user intention based on a pre-constructed matching model between user intention and service quality target.

[0157] For example, in some embodiments of the present application, the result of the networked business system executing the target networked business includes service level actual values of the plurality of service level indicators, and the apparatus 50 further includes:

[0158] The third processing unit is configured to determine whether a difference between a service level actual value corresponding to each service level indicator and a service level target value corresponding to each service level indicator is within a preset range.

[0159] a fourth processing unit, configured to determine a service adjustment strategy corresponding to the service level indicator in response to any of the difference values not being within the preset range;

[0160] a control unit, configured to control the network service system to execute the target network service based on the service adjustment strategy until the difference value between the actual value of the service level and the target value of the service level is within the preset range.

[0161] For example, in some embodiments of the present application, the device 50 further comprises:

[0162] a generating unit, configured to generate alarm information of the service level indicator exceeding the preset range of the difference value in response to the difference value not being within the preset range.

[0163] For example, in some embodiments of the present application, the second processing unit 504 is configured to determine a target network service strategy based on the target values of the plurality of service level indicators, including:

[0164] determining at least one network service strategy based on the target values of the plurality of service level indicators;

[0165] obtaining historical values of the plurality of service level indicators;

[0166] inputting the historical values of the plurality of service level indicators and the at least one network service strategy into a machine learning prediction algorithm model to obtain predicted values of the plurality of service level indicators and network resource state indicators corresponding to each network service strategy;

[0167] determining the target network service strategy from the at least one network service strategy based on the target values of the plurality of service level indicators, the predicted values of the plurality of service level indicators, and the network resource state indicators corresponding to each network service strategy.

[0168] For example, in some embodiments of the present application, the second processing unit 504 is configured to determine a plurality of network service strategies based on the target values of the plurality of service level indicators, including:

[0169] obtaining system function indicators and system performance indicators of the network service system;

[0170] determining the network service strategies based on the target values of the plurality of service level indicators, the system function indicators, and the system performance indicators.

[0171] For example, in some embodiments of this application, after executing the target computing network service based on the target computing network service policy, the device 50 further includes:

[0172] The acquisition unit is used to acquire the service level update values ​​of the multiple service level indicators and the network resource update status of the computing network service system;

[0173] The fifth processing unit is used to determine the evaluation value of the target computing network service strategy based on the service level update values ​​of the multiple service level indicators and the network resource update status of the computing network service system.

[0174] The computing network service execution device 50 provided in some embodiments of this application can execute the technical solution of the computing network service execution method in any of the above embodiments. Its implementation principle and beneficial effects are similar to those of the computing network service execution method. Please refer to the implementation principle and beneficial effects of the computing network service execution method. It will not be repeated here.

[0175] Figure 6 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of this application, such as... Figure 6 As shown, the electronic device may include: a processor 610, a communications interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communications interface 620, and the memory 630 communicate with each other through the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute the aforementioned computing network service execution method. The method includes: receiving service demand information from a user for a target computing network service, and extracting service demand feature information including Quality of User Experience (QoE) as a user intent based on the service demand information; determining the service intent corresponding to the user intent, wherein the service intent includes service level target values ​​of multiple service level indicators corresponding to the user intent; and determining a target computing network service service strategy based on the service level target values ​​of the multiple service level indicators; wherein the target computing network service service strategy is used by the computing network service system to execute the target computing network service.

[0176] Further, the logic instructions in the memory 630 described above can be implemented in the form of software functional units and sold or used as standalone products, which can be stored in a computer readable storage medium. Based on such an understanding, the technical solutions of the present application essentially or the parts that make contributions to the prior art or parts of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0177] In another aspect, the present application also provides a computer program product, which comprises a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program is executed by a processor, so that the computer can execute the above-mentioned algorithm network service execution method, which comprises: receiving service demand information of a user for a target algorithm network service, and extracting service demand characteristic information containing user experience quality QoE as a user intention based on the service demand information; determining a service intention corresponding to the user intention, the service intention comprising service level target values of a plurality of service level indicators corresponding to the user intention; determining a target algorithm network service service policy based on the service level target values of the plurality of service level indicators; wherein the target algorithm network service service policy is used for an algorithm network service system to execute the target algorithm network service.

[0178] In another aspect, the present application also provides a computer program product, which comprises a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program is executed by a processor, so that the computer can execute the above-mentioned algorithm network service execution method, which comprises: receiving service demand information of a user for a target algorithm network service, and extracting service demand characteristic information containing user experience quality QoE as a user intention based on the service demand information; determining a service intention corresponding to the user intention, the service intention comprising service level target values of a plurality of service level indicators corresponding to the user intention; determining a target algorithm network service service policy based on the service level target values of the plurality of service level indicators; wherein the target algorithm network service service policy is used for an algorithm network service system to execute the target algorithm network service.

[0179] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0180] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and the necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of software products, and the computer software products can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and include a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0181] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for executing a network service, characterized by, The method comprises: receiving service demand information of a user for a target network service, and extracting service demand characteristic information containing user experience quality as a user intention based on the service demand information; determining a service intention corresponding to the user intention, the service intention comprising service level target values of a plurality of service level indicators corresponding to the user intention; determining a target network service service strategy based on the service level target values of the plurality of service level indicators, wherein the target network service service strategy is used for a network service system to execute the target network service.

2. The method of claim 1, wherein, The method further comprises: inputting the service demand information into a machine learning model, performing inference on the service demand information, and obtaining an inference result; determining service demand characteristic information containing user experience quality in the inference result as the user intention.

3. The method according to claim 1 or 2, characterized in that, The method further comprises: translating the service demand information into a network configuration language or a programming language; extracting service demand characteristic information containing user experience quality as the user intention based on the network configuration language or the programming language.

4. The method of claim 1, wherein, The method further comprises: determining the service intention corresponding to the user intention based on a pre-constructed matching model between user intentions and service quality targets.

5. The method according to claim 1 or 2, characterized in that, The result of the network service system executing the target network service comprises service level actual values of the plurality of service level indicators, and the method further comprises: determining whether a difference between the service level actual values corresponding to each service level indicator and the service level target values corresponding to each service level indicator is within a preset range; in response to any of the differences not being within the preset range, determining a service adjustment strategy corresponding to the service level indicator; and controlling the network service system to execute the target network service based on the service adjustment strategy until the difference between the service level actual values and the service level target values is within the preset range.

6. The method of claim 5, wherein, The method further comprises: in response to the difference not being within the preset range, generating alarm information corresponding to the service level indicator that exceeds the preset range of the difference.

7. The method according to claim 1 or 2, characterized in that, The method further comprises: determining at least one network service service strategy based on the service level target values of the plurality of service level indicators; obtaining service level historical values of the plurality of service level indicators; inputting the service level historical values of the plurality of service level indicators and the at least one network service service strategy into a machine learning prediction algorithm model to obtain service level prediction values of the plurality of service level indicators and network resource state indicators corresponding to each network service service strategy; determining the target network service service strategy from the at least one network service service strategy based on the service level target values of the plurality of service level indicators, the service level prediction values, and the network resource state indicators corresponding to each network service service strategy.

8. The method of claim 7, wherein, The method further comprises: obtaining system function indicators and system performance indicators of the algorithm network service system; determining the algorithm network service service strategy based on the service level target values of the plurality of service level indicators, the system function indicators and the system performance indicators.

9. The method of claim 1 or 2, wherein, After executing the target algorithm network service based on the target algorithm network service service strategy, the method further comprises: obtaining service level update values of the plurality of service level indicators and network resource update states of the algorithm network service system; and determining an evaluation value of the target algorithm network service service strategy based on the service level update values of the plurality of service level indicators and the network resource update states of the algorithm network service system.

10. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the algorithm network service execution method of any one of claims 1 to 9. 11.A non-transitory computer-readable storage medium having stored thereon a computer program. The computer program is executed by the processor to implement the algorithm network service execution method of any one of claims 1 to 9.

12. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the algorithm network service execution method of any one of claims 1 to 9.

Images