CN113312556A - Cloud service mining method based on 5G interconnection big data and big data pushing system - Google Patents

Abstract

The embodiment of the disclosure provides a cloud service mining method and a big data pushing system based on 5G interconnection big data, in the process of carrying out business mining on the business feedback data, by introducing the social relation influence degree for representing the tendency behaviors of a plurality of social users to indicate the target cloud business influenced by the target knowledge entity in the business feedback data, namely, the various tendency information of a plurality of social users to the service feedback data is converted, then the mining cloud service information of the service feedback data is determined according to the target social influence characteristics, updating the information push weight of the target release cloud service in the service information release strategy based on the mining cloud service information, the method and the device enable the trend information corresponding to the social users to be merged into the mined cloud service information, and ensure the accuracy of the mining of the cloud service information, thereby improving the accuracy of the mining of the user service data and the pushing of the service information.

Description

Cloud service mining method based on 5G interconnection big data and big data pushing system

Technical Field

The disclosure relates to the technical field of big data, in particular to a cloud service mining method and a big data pushing system based on 5G interconnected big data.

Background

With the development of cloud computing technology, internet service providers have a large amount of online data, and with the arrival of the 5G interconnection era, the data amount is still rapidly increasing, and besides improving own services by using big data, the internet service providers have started to realize data commercialization and discover new commercial values by using the big data.

All the behaviors of the user are marked on the Internet platform, so that the Internet service provider can conveniently obtain a large amount of user behavior information. The information generated by the internet platform generally has authenticity and certainty, and the analysis of the data by using a big data technology can help the internet service provider to formulate a targeted information delivery service strategy, so that greater benefit is obtained.

However, in the current big data mining process, the accuracy of user service data mining and service information pushing, namely the accuracy of the user service data mining and the service information pushing, cannot reach the estimation effect.

Disclosure of Invention

In order to overcome at least the above disadvantages in the prior art, the present disclosure aims to provide a cloud service mining method and a big data pushing system based on 5G interconnected big data.

In a first aspect, the present disclosure provides a cloud service mining method based on 5G interconnected big data, which is applied to a big data push system, where the big data push system is in communication connection with a plurality of 5G internet terminals, and the method includes:

generating corresponding cloud service delivery data according to a service information delivery strategy determined based on the service delivery service interest points of the service registered users of the 5G internet terminal, and acquiring service feedback data of the service registered users on the cloud service delivery data;

performing feature extraction on the service feedback data according to preset social relationship features to obtain target social influence features, wherein the service feedback data comprise target knowledge entities, the preset social relationship features comprise a plurality of social relationship influence degrees, and each social relationship influence degree is used for representing the confidence degree that a social user tends to a cloud service where any knowledge entity is located in any user service data;

and determining mining cloud service information of the service feedback data according to the target social influence characteristics, and updating the information push weight of the target delivery cloud service in the service information delivery strategy based on the mining cloud service information.

In a second aspect, the embodiment of the disclosure further provides a cloud service mining system based on 5G interconnected big data, where the cloud service mining system based on 5G interconnected big data includes a big data pushing system and a plurality of 5G internet terminals in communication connection with the big data pushing system;

the big data push system is used for:

generating corresponding cloud service delivery data according to a service information delivery strategy determined based on the service delivery service interest points of the service registered users of the 5G internet terminal, and acquiring service feedback data of the service registered users on the cloud service delivery data;

performing feature extraction on the service feedback data according to preset social relationship features to obtain target social influence features, wherein the service feedback data comprise target knowledge entities, the preset social relationship features comprise a plurality of social relationship influence degrees, and each social relationship influence degree is used for representing the confidence degree that a social user tends to a cloud service where any knowledge entity is located in any user service data;

and determining mining cloud service information of the service feedback data according to the target social influence characteristics, and updating the information push weight of the target delivery cloud service in the service information delivery strategy based on the mining cloud service information.

According to any one of the above aspects, the present disclosure provides an implementation manner in which, during the process of performing service mining on the service feedback data, by introducing a social relationship influence degree for representing tendency behavior of a plurality of social users, to indicate target cloud traffic affected by a target knowledge entity in the traffic feedback data, namely, the various tendency information of a plurality of social users to the service feedback data is converted, then the mining cloud service information of the service feedback data is determined according to the target social influence characteristics, updating the information push weight of the target release cloud service in the service information release strategy based on the mining cloud service information, the method and the device enable the trend information corresponding to the social users to be merged into the mined cloud service information, and ensure the accuracy of the mining of the cloud service information, thereby improving the accuracy of the mining of the user service data and the pushing of the service information.

Drawings

Fig. 1 is an application scenario diagram of a cloud service mining system based on 5G interconnected big data according to an embodiment of the present disclosure;

fig. 2 is a schematic flow chart of a cloud service mining method based on 5G interconnected big data according to an embodiment of the present disclosure;

fig. 3 is a functional module schematic diagram of a cloud service mining device based on 5G interconnected big data according to an embodiment of the present disclosure;

fig. 4 is a block diagram schematically illustrating a structure of a big data pushing system for implementing the cloud service mining method based on 5G interconnected big data according to the embodiment of the present disclosure.

Detailed Description

The following describes in detail aspects of embodiments of the present disclosure with reference to the drawings attached hereto. In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular system structures, interfaces, techniques, etc. in order to provide a thorough understanding of the particular embodiments of the disclosure.

Fig. 1 is a scene schematic diagram of a 5G interconnection big data-based cloud service mining system 10 according to an embodiment of the present disclosure. The cloud service mining system 10 based on the 5G interconnected big data may include a big data push system 100 and a 5G interconnected internet terminal 200 communicatively connected to the big data push system 100. The 5G interconnection big data based cloud service mining system 10 shown in fig. 1 is only one possible example, and in other possible embodiments, the 5G interconnection big data based cloud service mining system 10 may also include only at least part of the components shown in fig. 1 or may also include other components.

In this embodiment, the big data pushing system 100 and the 5G internet terminal 200 in the cloud service mining system 10 based on 5G interconnected big data may cooperatively execute the cloud service mining method based on 5G interconnected big data described in the following method embodiment, and the detailed description of the following method embodiment may be referred to in the execution step parts of the big data pushing system 100 and the 5G internet terminal 200.

In order to solve the technical problem in the foregoing background art, fig. 2 is a schematic flow chart of a cloud service mining method based on 5G interconnection big data according to an embodiment of the present disclosure, where the cloud service mining method based on 5G interconnection big data according to the embodiment of the present disclosure may be executed by the big data push system 100 shown in fig. 1, and the cloud service mining method based on 5G interconnection big data is described in detail below.

Step S110, generating corresponding cloud service delivery data according to a service information delivery strategy determined by a service delivery interest point of a service registered user based on a 5G internet terminal, and acquiring service feedback data of the service registered user on the cloud service delivery data.

In this embodiment, the service delivery interest points of the service registered users may be determined according to the user behavior big data and the scene interaction behavior big data respectively crawled by the user behavior crawling program and the scene interaction behavior crawling program of the service registered users, which will be described later specifically.

In this embodiment, the cloud service delivery data matched with the service delivery policy rules of each delivery service indicated in the service information delivery policy can be acquired and pushed to the 5G internet terminal, and then, in the service use process of the service registered user, the service feedback data of the service registered user on the cloud service delivery data can be acquired in real time, and the service feedback data can represent the tendency intention of the service registered user for the cloud service delivery data, such as continuous attention behavior, sharing behavior and the like.

And step S120, extracting the characteristics of the service feedback data according to the preset social relationship characteristics to obtain target social influence characteristics.

The service feedback data comprises a target knowledge entity. For example, in an intelligent online education scenario, the business feedback data is online learning behavior business data, and the target knowledge entity is a certain online learning knowledge point. The preset social relation characteristics comprise a plurality of social relation influence degrees, each social relation influence degree is used for representing the confidence degree that one social user tends to the cloud service where any knowledge entity is located in any user service data, and the target social influence characteristics are used for representing the characteristic information contained in the service feedback data.

Step S130, mining cloud service information of the service feedback data is determined according to the target social influence characteristics, and information push weights of the target delivery cloud services in the service information delivery strategy are updated based on the mining cloud service information.

For example, the mining cloud service information may include tendency information that the service feedback data of each mining cloud service is concerned, and then update rule information of the corresponding information push weight may be acquired based on the tendency information that the service feedback data of each mining cloud service is concerned, and then the information push weight of the target delivery cloud service in the service information delivery policy is updated based on the update rule information of the corresponding information push weight.

Based on the above steps, in the process of service mining on the service feedback data, the embodiment indicates the target cloud service influenced by the target knowledge entity in the service feedback data by introducing the social relationship influence degree for representing the tendency behaviors of the plurality of social users, that is, converts the plurality of tendency information of the plurality of social users on the service feedback data, and then determines the mining cloud service information of the service feedback data according to the target social influence characteristic, so as to update the information push weight of the target delivery cloud service in the service information delivery strategy based on the mining cloud service information, so that the tendency information corresponding to the plurality of social users is merged into the mining cloud service information, thereby ensuring the accuracy of mining cloud service information, and improving the accuracy of user service data mining and service information pushing.

In one embodiment, step S130 can be implemented by the following exemplary sub-steps, which are described in detail below.

Step S131, feature reduction is carried out on the target social influence features, and first mining cloud service information of the service feedback data is obtained.

The first mining cloud service information is used for indicating a target cloud service influenced by the target knowledge entity in the service feedback data. The target social influence characteristics comprise characteristic information of the service feedback data, the preset social relation characteristics are blended into the target social influence characteristics, and the first mining cloud service information obtained by adopting a characteristic reduction mode is equivalent to mining cloud service information obtained by blending tendency information of the social users corresponding to the plurality of social relation influence degrees.

Step S132, according to the preset social relationship characteristics, performing user service data conversion on the first mining cloud service information to obtain a plurality of pieces of benchmark mining cloud service information.

Each datum mining cloud service information corresponds to a social relationship influence degree, each datum mining cloud service information is used for indicating target cloud services which are inclined by corresponding social users, and the target cloud services indicated by the datum mining cloud service information may have differences due to different inclination conditions of different social users. The first mining cloud service information is equivalent to mining cloud service information obtained after trend information of a plurality of social contact users is fused, and benchmark mining cloud service information matched with the plurality of social contact users is converted through the influence degree of the plurality of social contact relationships in the preset social contact relationship characteristic and the first mining cloud service information so as to indicate target cloud services influenced by the target knowledge entity in the service feedback data, so that mining cloud service information of the service feedback data can be updated on the basis of the converted mining cloud service information of the plurality of benchmarks, and the accuracy of mining cloud service information is improved.

Step S133, processing the target social influence characteristics based on the plurality of benchmark mining cloud service information to obtain second mining cloud service information of the service feedback data.

Because each datum mining cloud service information corresponds to a social relationship influence degree, and each datum mining cloud service information represents tendency information of a corresponding social user concerning the service feedback data, the target social influence characteristics are processed based on the plurality of datum mining cloud service information, so that the accuracy of the second mining cloud service information is improved.

In the process of service mining on the service feedback data, the reference mining cloud service information matched with the plurality of social users is converted by introducing the social relationship influence degree for expressing the tendency behaviors of the plurality of social users to indicate the target cloud service influenced by the target knowledge entity in the service feedback data, namely, the plurality of tendency information of the plurality of social users on the service feedback data is converted, and then the second mining cloud service information of the service feedback data is obtained through the plurality of reference mining cloud service information and the target social influence characteristics of the service feedback data, so that the tendency information corresponding to the plurality of social users is merged into the second mining cloud service information, the accuracy of the second mining cloud service information is ensured, and the accuracy of user service data mining is improved.

The following describes a flow of another cloud service mining method based on 5G interconnected big data provided by the embodiments of the present disclosure, and the method includes the following steps.

Step S210, performing feature extraction on the service feedback data to obtain a first social influence feature of the service feedback data.

The first social influence feature is used to represent feature information included in the service feedback data, the service feedback data includes a target knowledge entity, the service feedback data is arbitrary user service data, for example, in an intelligent online education scene, the service feedback data is online learning behavior service data, and the target knowledge entity is a certain online learning knowledge point.

In one embodiment, the step S210 includes: and calling a first feature extraction unit in the first cloud service mining network, and performing feature extraction on the service feedback data to obtain a first social influence feature of the service feedback data.

The first cloud service mining network is used for acquiring a model for mining cloud service information of the service feedback data, and for example, the first cloud service mining network is a long memory cycle network. The first feature extraction unit is configured to obtain social influence features of the service feedback data, for example, the first feature extraction unit is a feature extractor in a long memory cycle network. Feature extraction is performed on the service feedback data through a first feature extraction unit in the first cloud service mining network, so that the first social influence feature comprises feature information of the service feedback data, and accuracy of the obtained first social influence feature is guaranteed.

In one embodiment, the first feature extraction unit includes a plurality of first deep feature extraction nodes, and the process of obtaining the first social influence feature includes: calling a first depth feature extraction node according to the arrangement sequence of the plurality of first depth feature extraction nodes, performing feature extraction on the service feedback data to obtain a first reference social influence feature, calling the current first depth feature extraction node to perform feature extraction on the reference social influence feature output by the last first depth feature extraction node, obtaining a first reference social influence feature corresponding to the current first depth feature extraction node until the last first reference social influence feature output by the last first depth feature extraction node is obtained, and determining the first reference social influence feature output by the last first depth feature extraction node as the first social influence feature.

In the plurality of first depth feature extraction nodes included in the first feature extraction unit, the feature range of the reference social influence feature output by the plurality of first depth feature extraction nodes is gradually reduced according to the arrangement sequence of the plurality of first depth feature extraction nodes. The service feedback data are subjected to feature extraction according to a plurality of scales through a plurality of first depth feature extraction nodes in the feature extraction unit, namely the feature extraction unit gradually enriches the features contained in the social influence features in a downsampling mode so as to improve the accuracy of the first social influence features.

Step S220, fusing the preset social relationship characteristic with the first social influence characteristic to obtain a second social influence characteristic.

The preset social relationship characteristics comprise a plurality of social relationship influence degrees, each social relationship influence degree is used for representing the confidence degree that one social user tends to the cloud service where any knowledge entity is located in any user service data, and the social relationship influence degree can reflect the tendency condition of the corresponding social user. Because the tendency conditions of the plurality of social users are different, the influence degree of the social relationship corresponding to each social user is also divided into a plurality of degrees, the larger the influence degree of the social relationship is, the higher the tendency condition of the corresponding social user is, namely, the higher the confidence degree of the cloud service of the knowledge entity marked in the user service data by the social user is; the smaller the influence degree of the social relationship is, the lower the tendency condition of the corresponding social user is, that is, the lower the confidence degree of the cloud service of the social user in which the knowledge entity is marked in the user service data is.

In an embodiment, the preset social relationship characteristic is set arbitrarily, for example, the preset social relationship characteristic is [0.1, 01, 0.4, 0.4], that is, the preset social relationship characteristic includes social relationship influence degrees corresponding to 4 social users, the social relationship influence degree corresponding to two social users is 0.1, and the social relationship influence degree corresponding to two social users is 0.4.

Through fusing the preset social relationship characteristic with the first social relationship influence characteristic, the second social relationship influence characteristic not only includes the characteristic in the service feedback data, but also integrates the social relationship influence degree corresponding to a plurality of social users, so that the characteristic contained in the second social relationship influence characteristic is dynamically associated with the preset social relationship characteristic, and the characteristic contained in the second social relationship influence characteristic is influenced by the preset social relationship characteristic, the rich expression capability of the characteristic contained in the second social relationship influence characteristic is enriched, and the accuracy of the characteristic contained in the second social relationship influence characteristic is improved.

For example, in one embodiment, the step S220 includes: and calling a decision node in the first cloud service mining network, and fusing the preset social relation characteristic and the first social influence characteristic to obtain a second social influence characteristic.

The decision node is used for fusing the preset social relationship characteristic with the first social influence characteristic. Through the decision node, the preset social relationship characteristic and the first social influence characteristic of the service feedback data are fused, and the rich expression capability of the characteristics contained in the second social influence characteristic is enriched. For example, the feature range of the preset social relationship feature is expanded, so that the feature range of the preset social relationship feature after the feature range expansion is the same as that of the first social influence feature, and then the expanded preset social relationship feature is fused with the first social influence feature through the long-short term attention model, so that the features included in the first social influence feature are enriched, and the fused second social influence feature is obtained.

And step S230, performing feature restoration on the second social influence feature to obtain a target social influence feature.

Wherein, the target social influence characteristics are used for representing characteristic information contained in the business feedback data. After the second social influence feature is obtained, the feature included in the social influence feature is refined in a feature reduction mode, and accuracy of the feature included in the target social influence feature is improved.

In one embodiment, the step S230 includes: and calling a first feature restoration unit in the first cloud service mining network, and performing feature restoration on the second social influence feature to obtain the target social influence feature.

The first feature restoring unit is used for enriching features contained in the social influence features, for example, the first feature restoring unit is a feature restorer in a long memory cycle network.

In one embodiment, the first cloud service mining network includes a first feature extraction unit, the first feature extraction unit includes a plurality of first depth feature extraction nodes, the first feature restoration unit includes a plurality of second depth feature extraction nodes, and the process of obtaining the target social influence feature includes: calling a first second depth feature extraction node according to the arrangement sequence of a plurality of second depth feature extraction nodes, performing feature reduction on the second social influence feature to obtain a first second reference social influence feature, calling a current second depth feature extraction node, performing feature reduction on the reference social influence feature output by the last second depth feature extraction node and the reference social influence feature output by the first depth feature extraction node corresponding to the current second depth feature extraction node to obtain a second reference social influence feature corresponding to the current second depth feature extraction node until the last second reference social influence feature output by the second depth feature extraction node is obtained, and determining the second reference social influence feature output by the last second depth feature extraction node as the target social influence feature.

In the plurality of second depth feature extraction nodes included in the first feature restoration unit, the feature range of the reference social influence feature output by the plurality of second depth feature extraction nodes is gradually increased according to the arrangement sequence of the plurality of second depth feature extraction nodes. And gradually refining the features contained in the social influence features by adopting an up-sampling mode through a plurality of second depth feature extraction nodes in the feature restoration unit so as to improve the accuracy of the features contained in the target social influence features.

It should be noted that, in the embodiment of the present disclosure, after the first social influence feature extracted from the service feedback data feature is fused with the preset social relationship feature, the obtained second social influence feature is subjected to feature restoration to obtain the target social influence feature, but in another embodiment, the above steps S210 to S230 are not required to be performed, and other ways may be adopted to perform feature extraction on the service feedback data according to the preset social relationship feature to obtain the target social influence feature.

In one embodiment, a first cloud service mining network is called, and service feedback data are subjected to feature extraction according to preset social relation features to obtain target social influence features. And acquiring the target social influence characteristics through the first cloud service mining network so as to improve the accuracy of the target social influence characteristics.

Step S240, feature reduction is carried out on the target social influence features, and first mining cloud service information of the service feedback data is obtained.

The first mining cloud service information is used for indicating a target cloud service influenced by the target knowledge entity in the service feedback data. In an embodiment, the first mining cloud service information includes distribution probabilities corresponding to a plurality of service feedback objects in the service feedback data, where the distribution probabilities are used to represent probability values of the corresponding service feedback objects in the target cloud service. The first mining cloud service information comprises a plurality of service feedback objects, the tendency value of each service feedback object represents the distribution probability corresponding to the service feedback object located at the same service node in the service feedback data, and for the service feedback object of any service node in the service feedback data, the tendency value of the service feedback object located at the same service node in the first mining cloud service information is the distribution probability of the service feedback object of the service node in the service feedback data.

In an embodiment, the first mining cloud service information is represented in a form of a probability distribution space, and in the first mining cloud service information, the greater the distribution probability corresponding to a service feedback object is, the deeper the probability depth corresponding to the service feedback object is, the smaller the distribution probability corresponding to the service feedback object is, and the shallower the probability depth corresponding to the service feedback object is.

The target social influence characteristics not only comprise characteristic information of the service feedback data, but also incorporate preset social relation characteristics, and the first mining cloud service information is obtained by adopting a characteristic reduction mode for the target social influence characteristics, wherein the first mining cloud service information is equivalent to mining cloud service information obtained by fusing tendency information of social users corresponding to a plurality of social relation influence degrees.

In one embodiment, the step S240 includes: and calling a first cloud service mining network, and performing feature restoration on the target social influence features to obtain first mining cloud service information of the service feedback data.

In one embodiment, a deep feature extraction node in the first cloud service mining network is called, and deep feature extraction processing is performed on the target social influence feature to obtain the first mining cloud service information.

And step S250, splicing the service feedback data and the first mining cloud service information to obtain spliced user service data.

The first mining cloud service information is used for indicating a target cloud service influenced by the target knowledge entity in the service feedback data, and the service feedback data is spliced with the first mining cloud service information, so that the spliced user service data not only contains information contained in the service feedback data, but also contains information used for indicating the target cloud service influenced by the target knowledge entity in the service feedback data, and the information contained in the spliced user service data is enriched, so that a plurality of pieces of standard mining cloud service information can be converted in the subsequent process.

And step S260, performing feature extraction on the spliced user service data to obtain a third social influence feature.

The third social influence characteristic is used for representing characteristic information contained in the business feedback data and indicating characteristic information of the target cloud business influenced by the target knowledge entity in the business feedback data.

In one embodiment, the step S260 includes: and calling a feature extraction unit in the user service data conversion model, and performing feature extraction on the spliced user service data to obtain the third social influence feature.

The user service data conversion model is used for converting benchmark mining cloud service information corresponding to a plurality of social relation influence degrees, and the feature extraction unit is used for acquiring social influence features of spliced user service data, and is a feature extractor in a long memory cycle network. The feature extraction unit in the user service data conversion model is the same as the first feature extraction unit in the first cloud service mining network in step S210, and is not described herein again.

And step S270, fusing the preset social relationship characteristic and the third social influence characteristic to obtain a fourth social influence characteristic.

By fusing the preset social relationship characteristic with the third social relationship influence characteristic, the fourth social relationship influence characteristic not only includes characteristic information contained in the service feedback data, but also is fused with social relationship influence degrees corresponding to a plurality of social users, so that the reference mining cloud service information corresponding to each social relationship influence degree can be converted according to the fourth social relationship influence characteristic in the following process.

In one embodiment, the step S270 includes: and calling a feature fusion unit in the user service data conversion model, and fusing the preset social relationship feature and the third social influence feature to obtain a fourth social influence feature.

The feature fusion unit is similar to the decision node in step S220, and is not described herein again.

And step S280, performing feature restoration on the fourth social influence feature to obtain a plurality of datum mining cloud service information.

Each datum mining cloud service information corresponds to a social relationship influence degree, each datum mining cloud service information is used for indicating target cloud services which are inclined by corresponding social users, and the target cloud services indicated by the datum mining cloud service information may have differences due to different inclination conditions of different social users.

In an embodiment, each datum mining cloud service information includes first distribution probabilities corresponding to a plurality of service feedback objects in service feedback data, where the first distribution probabilities are used to represent probability values of the corresponding service feedback objects in a target cloud service, and then, by marking the plurality of first distribution probabilities included in user service data, a target cloud service influenced by a target knowledge entity to which a corresponding social user tends in the service feedback data can be determined. The method comprises the steps that any datum mining cloud service information comprises a plurality of service feedback objects, the tendency value of each service feedback object is the first distribution probability contained in the datum mining cloud service information, and for the service feedback object of any service node in the service feedback data, the tendency value of the service feedback object positioned at the same service node in the datum mining cloud service information is the first distribution probability of the service feedback object of the service node in the service feedback data.

In the embodiment of the disclosure, the fourth social influence feature includes not only feature information of the first mined cloud service information and feature information of the service feedback data, but also a preset social relationship feature, and the first mined cloud service information is equivalent to mined cloud service information obtained by fusing tendency information of social users corresponding to a plurality of social relationship influence degrees, and the fourth social influence feature is processed by adopting a feature reduction mode, so that benchmark mined cloud service information corresponding to the plurality of social relationship influence degrees can be converted, tendency information of the plurality of social users to the service feedback data is reduced, that is, benchmark mined cloud service information corresponding to each social user, so as to update the mined cloud service information of the service feedback data later.

In one embodiment, the step S280 includes: and calling a feature reduction unit in the user service data conversion model, and carrying out feature reduction on the fourth social influence feature to obtain a plurality of datum mining cloud service information.

The feature restoring unit is similar to the first feature restoring unit in the first cloud service mining network in step S230, and is not described herein again.

It should be noted that in the embodiment of the present disclosure, the service feedback data is introduced, and the service feedback data and the preset social relationship characteristic are fused to convert the plurality of pieces of benchmark mining cloud service information, but in another embodiment, the above-mentioned steps S250 to S280 are not required to be performed, and other manners may be adopted to perform user service data conversion on the first mining cloud service information according to the preset social relationship characteristic, so as to obtain the plurality of pieces of benchmark mining cloud service information.

In one embodiment, a user service data conversion model is called, and user service data conversion is carried out on the first mining cloud service information according to the preset social relation characteristics to obtain a plurality of pieces of benchmark mining cloud service information. And converting the benchmark mining cloud service information corresponding to the influence degree of the social relationship according to the preset social relationship characteristic and the first mining cloud service information through a user service data conversion model so as to ensure the accuracy of the benchmark mining cloud service information.

Step S290, mining a change in a service operation behavior between cloud service information according to a plurality of benchmarks, and determining user service migration behavior data.

The user service migration behavior data is used for indicating service operation behavior change among target cloud services indicated by the plurality of datum mining cloud service information. Because each datum mining cloud service information corresponds to a social relationship influence degree, namely each datum mining cloud service information is equivalent to tendency information of a social user corresponding to the corresponding social relationship influence degree for paying attention to service feedback data, target cloud services indicated by the datum mining cloud service information are different due to different tendency conditions of a plurality of social users, user service migration behavior data can be determined through service operation behavior changes among the datum mining cloud service information, and the user service migration behavior data can indicate cloud services with mining values in the target cloud services inclined by the plurality of social users.

In one embodiment, the step S290 includes the following steps S291-S294.

Step S291, determining difference user service data between the plurality of pieces of benchmark mining cloud service information and election user service data of the plurality of pieces of benchmark mining cloud service information.

The cloud service data comprises balanced user service data of the first distribution probability of each service feedback object in the service feedback data in the plurality of reference mining cloud service information, the elected user service data can show the consistency between target cloud services indicated by the plurality of reference mining cloud service information, and each difference user service data comprises difference data between each first distribution probability in the corresponding reference mining cloud service information and the corresponding election in the elected user service data. The method comprises the steps of determining election user service data of a plurality of datum mining cloud service information, and then determining difference user service data between each datum mining cloud service information and the election user service data, so that a plurality of difference user service data are obtained.

In one embodiment, step S291 includes: determining election data of first distribution probabilities corresponding to service feedback objects located at the same service node in a plurality of pieces of benchmark mining cloud service information, constructing the differential user service data according to the plurality of obtained election data, determining differential data between the plurality of first distribution probabilities in the benchmark mining cloud service information and the corresponding election data in the differential user service data for each piece of benchmark mining cloud service information, and forming the plurality of obtained differential data into the differential user service data corresponding to the benchmark mining cloud service information.

Step S292 is to determine the sum of squares of the trend values of the service feedback objects located in the same service node in the multiple differentiated user service data.

The trend value of any service feedback object in any difference user service data is a difference value between the reference mining cloud service information corresponding to the difference user service data and a first distribution probability corresponding to a service feedback object of which the service feedback object is located at the same service node, and an election corresponding to the service feedback object of which the service feedback object is located at the same service node in election user service data.

For any service node, determining the square of the tendency value of the service feedback object positioned in the service node in the plurality of differential user service data, determining the sum of the squares of the tendency values corresponding to the plurality of differential user service data as the sum of the squares of the tendency values corresponding to the service node, and repeating the above manner to obtain the sum of the squares of the tendency values corresponding to the plurality of service nodes.

Step S293, determining the evolution of the ratio between the sum of squares corresponding to each service node and the number of the plurality of reference mining cloud service information as the second distribution probability of each service node, respectively.

The second distribution probability is used for representing the difference of tendency information of the corresponding service node in the plurality of datum mining cloud service information, and the tendency information represents whether a service feedback object located at the corresponding service node is in the target cloud service.

Step S294, constructing user service migration behavior data according to the second distribution probabilities of the plurality of service nodes.

The user service migration behavior data comprises second distribution probabilities corresponding to a plurality of service feedback objects in the service feedback data.

Step S310, fusing the target social influence characteristics with the user service migration behavior data to obtain second mining cloud service information.

Because the user service migration behavior data can indicate the service operation behavior change among the target cloud services indicated by the plurality of pieces of benchmark mining cloud service information, and the target social influence characteristics not only include the characteristic information of the service feedback data, but also fuse the preset social relation characteristics, uncertain cloud services in the plurality of pieces of benchmark mining cloud service information can be distinguished by fusing the target social influence characteristics and the user service migration behavior data, so that the accuracy of the target cloud services indicated by the second mining cloud service information is improved.

In one embodiment, this step S310 includes the following steps S3101-3105.

Step S3101, electing user service data of the plurality of reference mining cloud service information is determined.

This step is similar to the manner of determining the election of the user service data in step S2901, and is not described herein again.

Step S3102, determining the fusion feature of the target social influence feature and the user service migration behavior data and the weighting feature of the target social influence feature as a first fusion social influence feature.

The first fusion social influence feature is used for representing inconsistent information among benchmark mining cloud service information corresponding to a plurality of social relation influence degrees. Determining the fusion characteristics of the target social influence characteristics and the user service migration behavior data, and then determining the determined fusion characteristics and the weighting characteristics of the target social influence characteristics as the first fusion social influence characteristics. By obtaining the first fusion social influence feature, the features of the target social influence feature in the migration behavior indicated by the user service migration behavior data are enriched, so that the accuracy of the first fusion social influence feature is improved.

In one embodiment, step S3102 includes: determining fusion characteristics of the target social influence characteristics and tendency units of the user service migration behavior data, and determining the sum of the obtained fusion characteristics and the tendency units of the target social influence characteristics as the first fusion social influence characteristics. The tendency unit fusion feature refers to fusion feature of tendency values of the target social influence feature and the service feedback objects located at the same service node in the user service migration behavior data, and the sum of the tendency units refers to the sum of the obtained fusion feature and the tendency values of the service feedback objects located at the same service node in the target social influence feature.

Step S3103, determining the fusion feature of the target social influence feature and the election user business data and the weighting feature of the target social influence feature as a second fusion social influence feature.

The second fusion social influence feature is used for representing consistent information among the benchmark mining cloud service information corresponding to the social relation influence degrees. Determining a fusion feature of the target social influence feature and the election user business data, and then determining the determined fusion feature and a weighting feature of the target social influence feature as the second fusion social influence feature. By obtaining the second fusion social influence feature, the features in the cloud service, which are mapped into the target cloud service by the plurality of datum mining cloud service information in the target social influence feature, are enriched, and the accuracy of the second fusion feature is improved.

In one embodiment, step S3103 includes: and determining fusion characteristics of the target social influence characteristics and tendency units of the election user service data, and determining the sum of the obtained fusion characteristics and the tendency units of the target social influence characteristics as the second fusion social influence characteristics. The tendency unit fusion feature refers to the fusion feature of the tendency values of the target social influence feature and the service feedback objects located at the same service node in the election user service data, and the sum of the tendency units refers to the sum of the obtained fusion feature and the tendency values of the service feedback objects located at the same service node in the target social influence feature.

Step S3104, the first and second fused social influence features are spliced to obtain a spliced and fused social influence feature.

For example, the first fused social influence feature has a feature range of B × C1 × H × W, the second fused social influence feature has a feature range of B × C2 × H × W, and the feature range of the spliced fused social influence feature after splicing is B × C1+ C2 × H × W.

And 3105, performing deep feature extraction processing on the splicing fusion social influence features to obtain second mining cloud service information.

Since the splicing fusion social influence feature includes the rich features of the target social influence feature in the tendency service and the rich features of the migration behavior service, when the splicing fusion feature is subjected to deep feature extraction processing, the determined target cloud service in the splicing social influence feature can be distinguished from other areas, so that the accuracy of the target cloud service indicated by the second mining cloud service information is improved, that is, the accuracy of the second mining cloud service information is improved.

It should be noted that in the embodiment of the present disclosure, the user service migration behavior data is obtained first, and then the second mining cloud service information is obtained based on the target social influence feature and the user service migration behavior data, but in another embodiment, the above steps S290-S310 do not need to be executed, and other manners can be adopted to process the target social influence feature based on the plurality of pieces of benchmark mining cloud service information, so as to obtain the second mining cloud service information of the service feedback data.

In one embodiment, a second cloud service mining network is called, and the target social influence characteristics are processed based on a plurality of pieces of benchmark mining cloud service information to obtain second mining cloud service information of service feedback data. The second cloud service mining network is used for acquiring second mining cloud service information. And through the second cloud service mining network, the inconsistent information and the consistent information corresponding to the cloud service information are mined by using a plurality of benchmarks so as to ensure the accuracy of the second cloud service information.

In the method provided by the embodiment of the disclosure, in the process of service mining on the service feedback data, the reference mining cloud service information matched with the plurality of social users is converted by introducing the social relationship influence degree for representing the tendency behaviors of the plurality of social users to indicate the target cloud service influenced by the target knowledge entity in the service feedback data, that is, the plurality of tendency information of the plurality of social users on the service feedback data is converted, and then the second mining cloud service information of the service feedback data is acquired by the plurality of reference mining cloud service information and the target social influence characteristics of the service feedback data, so that the tendency information corresponding to the plurality of social users is merged into the second mining cloud service information, the accuracy of the second mining cloud service information is ensured, and the accuracy of mining the user service data is improved.

In addition, through a decision node in the first mining cloud service information model and a feature fusion unit in the user service data conversion model, in the process of acquiring mining cloud service information of service feedback data, a preset social relationship feature can be introduced, so that the preset social relationship feature is embedded into the feature fusion of the service feedback data, and the expression capacity of model decision is improved.

Based on the above embodiments, it can be seen that the process of performing user service data mining on the service feedback data can be performed by using the first cloud service mining network, the user service data conversion model, and the second cloud service mining network, and before the first cloud service mining network, the user service data conversion model, and the second cloud service mining network are called, the first cloud service mining network, the user service data conversion model, and the second cloud service mining network need to be trained, and the training process is described in detail in the following embodiments.

The embodiment of the disclosure also provides a flow of a cloud service mining method based on 5G interconnection big data, and the method comprises the following steps.

Step S310, acquiring reference service feedback data, a plurality of reference datum mining cloud service information and preset social relation characteristics.

The reference service feedback data comprises reference knowledge entities, each piece of reference datum mining cloud service information corresponds to a social relationship influence degree, each piece of reference datum mining cloud service information is used for indicating a reference cloud service influenced by a reference knowledge entity inclined by a corresponding social user in the reference service feedback data, and each piece of reference datum mining cloud service information is real inclination information of the corresponding social user on the reference service feedback data.

Step S320, calling a first cloud service mining network, and performing feature extraction on the reference service feedback data according to the preset social relationship features to obtain target reference social influence features. The steps are similar to the steps S210 to S230, and are not described herein again.

Step S330, a first cloud service mining network is called, the target reference social influence characteristics are subjected to characteristic restoration, and first reference mining cloud service information of the reference service feedback data is obtained. The first reference mining cloud service information is used for indicating reference cloud services influenced by the reference knowledge entity in the reference service feedback data. This step is similar to the step S240, and will not be described herein again.

Step S340, splicing the reference service feedback data and the first reference mining cloud service information to obtain first reference spliced user service data. This step is similar to the step S250, and will not be described herein again.

And step S350, calling a feature extraction unit to extract features of the first reference splicing user service data to obtain first reference social influence features. In the embodiment of the disclosure, the user service data conversion model comprises a feature extraction unit, a feature fusion unit and a feature restoration unit. The step is similar to the step S260, and is not described herein again.

And step S360, calling a feature fusion unit, and fusing the preset social relationship feature and the first reference social influence feature to obtain a second reference social influence feature. This step is similar to the step S270, and will not be described herein again.

Step S370, calling a feature restoration unit, and performing feature restoration on the second reference social influence feature to obtain a plurality of prediction benchmark mining cloud service information. And each piece of forecast benchmark mining cloud service information is corresponding to a social relationship influence degree and is used for indicating the forecast reference cloud service inclined by the corresponding social user. The step is similar to the step S280, and is not described herein again.

It should be noted that in the embodiment of the disclosure, reference service feedback data and first reference mining cloud service information are spliced, and then a feature extraction unit, a feature fusion unit and a feature restoration unit in a user service data conversion model are called to obtain a plurality of pieces of prediction benchmark mining cloud service information, while in another embodiment, steps S340 to S370 do not need to be executed, and a user service data conversion model can be called in other ways, and user service data conversion is performed on the first reference mining cloud service information according to a preset social relationship feature to obtain a plurality of pieces of prediction benchmark mining cloud service information.

Step S380, a second cloud service mining network is called, and the target reference social influence characteristics are processed on the basis of the plurality of prediction benchmark mining cloud service information, so that prediction mining cloud service information of the reference service feedback data is obtained. The steps are similar to the steps S290-S310, and are not described herein again.

Step 390, performing weighted fusion on the multiple pieces of reference mining cloud service information according to the preset social relationship characteristics to obtain fused reference mining cloud service information.

Because the preset social relationship characteristic comprises a plurality of social relationship influence degrees, the plurality of social relationship influence degrees correspond to the plurality of reference datum mining cloud service information one by one, the plurality of reference datum mining cloud service information is weighted and fused through the plurality of social relationship influence degrees in the preset social relationship characteristic, the obtained fusion datum mining cloud service information is used as a final result inclined by a plurality of social users, and therefore the fusion datum mining cloud service information is used as a supervision value in the follow-up process, and the first cloud service mining network, the user service data conversion model and the second cloud service mining network are trained.

In an embodiment, each reference mining cloud service information includes a distribution probability corresponding to a plurality of service feedback objects in the service feedback data, and then the step S390 includes: according to the preset social relationship characteristics, weighted fusion is carried out on the tendency values of the service feedback objects positioned at the same service node in the plurality of reference mining cloud service information, the fusion distribution probability of each service node is obtained, and the fusion reference mining cloud service information is converted according to the fusion weight of the plurality of service nodes. And mining the tendency value of the service feedback object of any service node in the cloud service information for any reference standard, namely the distribution probability corresponding to the service feedback object positioned at the same service node in the service feedback data.

Step S400, training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to service operation behavior changes between the forecast mining cloud service information and the fusion benchmark mining cloud service information.

Since the fused benchmark mining cloud service information is equivalent to the real benchmark mining cloud service information of the reference service feedback data, namely, the reference cloud service influenced by the reference knowledge entity indicated by the fusion benchmark mining cloud service information in the reference service feedback data is the real area of the reference knowledge entity in the reference service feedback data, the forecast mining cloud service information is forecasted by the first cloud service mining network, the user service data conversion model and the second cloud service mining network, based on the change of the business operation behavior between the predicted mining cloud business information and the fusion benchmark mining cloud business information, the distrust of the first cloud business mining network, the user business data conversion model and the second cloud business mining network can be determined, and the first cloud service mining network, the user service data conversion model and the second cloud service mining network are adjusted in the following process.

In one embodiment, the step S400 includes the following steps S401-S402.

Step S401, determining a first model evaluation index value according to business operation behavior change between the forecast excavation cloud business information and the fusion reference excavation cloud business information.

The first model evaluation index value is used for representing the change of business operation behaviors between the prediction segmentation and the fusion benchmark mining cloud business information, the larger the model evaluation index value is, the lower the confidence degrees of the first cloud business mining network, the user business data conversion model and the second cloud business mining network are represented, the smaller the model evaluation index value is, and the higher the confidence degrees of the first cloud business mining network, the user business data conversion model and the second cloud business mining network are represented.

And S402, training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the first model evaluation index value.

And training the first cloud service mining network, the user service data conversion model and the second cloud service mining network through the first model evaluation index value so as to reduce the first model evaluation index value and improve the confidence degrees of the first cloud service mining network, the user service data conversion model and the second cloud service mining network. For example, the forecast mining cloud service information is obtained through a first cloud service mining network, a user service data conversion model and a second cloud service mining network, wherein the decision node is a sub-model in the first cloud service mining network. The method comprises the steps of obtaining fusion benchmark mining cloud service information by weighting and fusing a plurality of pieces of reference benchmark mining cloud service information, then determining a first model evaluation index value according to service operation behavior change between the prediction mining cloud service information and the fusion benchmark mining cloud service information, and training a first cloud service mining network, a user service data conversion model and a second cloud service mining network according to the determined first model evaluation index value.

In one embodiment, the step S402 includes the following three ways.

The first mode is as follows: and determining a second model evaluation index value according to the change of the service operation behavior between the first reference mining cloud service information and the fusion benchmark mining cloud service information, and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the first model evaluation index value and the second model evaluation index value.

The second model evaluation index value is used for representing the change of the business operation behavior between the first reference mining cloud business information and the fusion benchmark mining cloud business information, the larger the change of the business operation behavior between the first reference mining cloud business information and the fusion benchmark mining cloud business information is, the larger the second model evaluation index value is, the smaller the change of the business operation behavior between the first reference mining cloud business information and the fusion benchmark mining cloud business information is, and the smaller the second model evaluation index value is.

And training the first cloud service mining network, the user service data conversion model and the second cloud service mining network through the first model evaluation index value and the second model evaluation index value so as to reduce the first model evaluation index value and the second model evaluation index value and improve the confidence degrees of the first cloud service mining network, the user service data conversion model and the second cloud service mining network.

In one embodiment, a process of training a first cloud service mining network, a user service data conversion model, and a second cloud service mining network according to a first model evaluation index value and a second model evaluation index value includes: and determining a first added value of the first model evaluation index value and the second model evaluation index value, and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the first added value.

The second mode is as follows: and determining a third model evaluation index value according to the service operation behavior change between the plurality of pieces of prediction benchmark mining cloud service information and the corresponding reference benchmark mining cloud service information, and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the first model evaluation index value and the third model evaluation index value.

The third model evaluation index value is conversion loss and is used for representing service operation behavior change between the plurality of pieces of prediction benchmark mining cloud service information and the corresponding reference benchmark mining cloud service information.

In one embodiment, a process of training a first cloud service mining network, a user service data conversion model, and a second cloud service mining network according to a first model evaluation index value and a third model evaluation index value includes: and determining a second added value of the first model evaluation index value and the third model evaluation index value, and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the second added value.

The third mode is as follows: the method comprises the steps of splicing reference business feedback data and fused benchmark mining cloud business information to obtain second reference spliced user business data, calling a feature extraction unit, carrying out feature extraction on the second reference spliced user business data to obtain a third reference social influence feature, determining a fourth model evaluation index value according to business operation behavior change between the third reference social influence feature and the first reference social influence feature, and training a first cloud business mining network, a user business data conversion model and a second cloud business mining network according to the first model evaluation index value and the fourth model evaluation index value.

And the fourth model evaluates the index value as consistency loss and is used for representing the business operation behavior change between the third reference social influence characteristic and the first reference social influence characteristic. The process of obtaining the second reference splicing user service data is the same as the step S250, and is not described herein again. The process of calling the feature extraction unit to obtain the third reference social influence feature is the same as the step S450, and is not described herein again.

Since the first reference social influence feature is a call to a feature extraction unit in the user business data transformation model, the reference service feedback data and the first reference spliced user service data spliced by the first reference mining cloud service information are subjected to feature extraction to obtain the reference service feedback data, the third reference social influence characteristic is obtained by calling the same characteristic extraction unit to extract the characteristics of the reference service feedback data and the second reference splicing user service data spliced by the fusion benchmark mining cloud service information, the first reference mining cloud service information is obtained by prediction, the fusion benchmark mining cloud service information is a real result of a plurality of social user labels, and evaluating the index value through a fourth model to determine the business operation behavior change between a first reference social influence characteristic corresponding to the prediction result output by the same characteristic extraction unit and a third reference social influence characteristic corresponding to the real result.

In one embodiment, a process of training a first cloud service mining network, a user service data conversion model, and a second cloud service mining network according to a first model evaluation index value and a fourth model evaluation index value includes: and determining a third added value of the first model evaluation index value and the fourth model evaluation index value, and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the third added value.

In one embodiment, the feature extraction unit includes a plurality of third depth feature extraction nodes, in the process of calling the plurality of third depth feature extraction nodes to perform feature extraction on the first reference mosaic user service data, a first third depth feature extraction node is called to perform feature extraction on the first reference mosaic user service data to obtain a first third reference social influence feature, a current third depth feature extraction node is called to perform feature extraction on a third reference social influence feature output by a last third depth feature extraction node to obtain a third reference social influence feature corresponding to the current third depth feature extraction node until a last third reference social influence feature output by the last third depth feature extraction node is obtained, and the third reference social influence feature output by the last third depth feature extraction node is determined as the first reference social influence feature. Similarly, in the process of calling a plurality of third depth feature extraction nodes to perform feature extraction on the second reference splicing user service data, according to the above manner, a plurality of fourth reference social influence features corresponding to the second reference splicing user service data can also be obtained, and a fourth reference social influence feature output by the last third depth feature extraction node is determined as the second reference splicing user service data.

It should be noted that, the above description is only made by training the first cloud service mining network, the user service data conversion model, and the second cloud service mining network in three ways, respectively, and in another embodiment, the three ways can be combined in pairs, for example, the first way is combined with the second way, and the second way is combined with the third way; alternatively, the above three ways are combined.

In one embodiment, the step S402 includes: determining a second model evaluation index value according to the business operation behavior change between the first reference mining cloud business information and the fusion benchmark mining cloud business information, determining a third model evaluation index value according to the business operation behavior change between the plurality of prediction benchmark mining cloud business information and the corresponding reference mining cloud business information, splicing the reference business feedback data and the fusion benchmark mining cloud business information to obtain second reference spliced user business data, calling a feature extraction unit to perform feature extraction on the second reference spliced user business data to obtain a third reference social influence feature, determining a fourth model evaluation index value according to the business operation behavior change between the third reference social influence feature and the first reference social influence feature, and determining the fourth model evaluation index value according to the first model evaluation index value, the second model evaluation index value, the third model evaluation index value and the fourth model evaluation index value, and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network.

In one embodiment, the process of training the first cloud service mining network, the user service data conversion model, and the second cloud service mining network according to the first model evaluation index value, the second model evaluation index value, the third model evaluation index value, and the fourth model evaluation index value includes: determining a total model evaluation index value according to the first model evaluation index value, the second model evaluation index value, the third model evaluation index value and the fourth model evaluation index value, and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the total model evaluation index value.

In the model training mode provided by the embodiment of the disclosure, in the process of service mining on the service feedback data, the reference mining cloud service information matched with the plurality of social users is converted by introducing the social relationship influence degree for representing the tendency behaviors of the plurality of social users to indicate the target cloud service influenced by the target knowledge entity in the service feedback data, that is, the plurality of tendency information of the plurality of social users on the service feedback data is converted, and then the second mining cloud service information of the service feedback data is acquired by the plurality of reference mining cloud service information and the target social influence characteristics of the service feedback data, so that the tendency information corresponding to the plurality of social users is merged into the second mining cloud service information, the accuracy of the second mining cloud service information is ensured, and the accuracy of mining the user service data is improved.

In addition, by considering the service operation behavior change between the converted multiple pieces of forecast benchmark mining cloud service information and the multiple pieces of reference benchmark mining cloud service information, the service operation behavior change between the third reference social influence characteristic and the first reference social influence characteristic, and the service operation behavior change between the first reference mining cloud service information and the fused benchmark mining cloud service information, the accuracy of the first cloud service mining network, the user service data conversion model and the second cloud service mining network is improved.

The model training method provided by the embodiment of the disclosure is used for training a model based on a plurality of reference bases to mine cloud service information. In the model training process of the related technology, the model can be trained by mining cloud service information by adopting the reference benchmark of a single social user. The accuracy of the model after model training is high by adopting the reference benchmarks of a plurality of social users to mine the cloud service information.

In addition, the benchmark mining cloud service information of a plurality of social users can be converted through the provided user service data conversion model, the correlation between the obtained forecast mining cloud service information and the reference benchmark mining cloud service information can be enriched, and the uncertainty between the social users can be estimated.

In one embodiment, for step S110, in the process of the service information delivery policy determined according to the service delivery service interest points of the service registered users based on the 5G internet terminal, the following exemplary steps may be implemented.

Step a110, obtaining user behavior big data and scene interaction behavior big data respectively crawled by a user behavior crawling program and a scene interaction behavior crawling program of a service registered user of the 5G internet terminal 200.

In order to avoid the problem of insufficient identification precision caused by identification based on single-dimension behavior big data to a certain extent, in the embodiment of the disclosure, the behavior big data of multiple dimensions can be crawled in multiple ways, and the behavior big data is synthesized to determine the service delivery interest point, so that the linkage consideration effect among the behavior big data of different dimensions is realized.

For example, first, a user behavior crawling program may crawl user behavior big data corresponding to a service registered user in a behavior operation process, where the user behavior big data is data capable of representing a user behavior state, and may include, for example, operation object data and operation flow data of a user behavior on each business plate; the scene interaction behavior crawling program can crawl scene interaction behavior big data corresponding to the service registered user in a behavior operation process, the scene interaction behavior big data are data capable of showing a scene interaction state in a scene interaction behavior mode, and the scene interaction behavior big data can be operation object data, operation flow data and the like of the scene interaction behavior on each service plate.

That is, in order to determine the service delivery interest point based on the multidimensional service delivery information, user behavior big data and scene interaction behavior big data respectively crawled by a user behavior crawling program and a scene interaction behavior crawling program of a service registered user can be acquired.

Step A120, determining first released service interest information corresponding to the target released service according to the user behavior big data, and determining second released service interest information of the target released service according to the scene interaction behavior big data.

Based on the acquired user behavior big data, determining first released service interest information corresponding to a target released service from the dimension of the behavior condition of the user, wherein the first released service interest information comprises subsequent released service interest points corresponding to the user behavior big data; based on the acquired scene interaction behavior big data, second released service interest information of the target released service can be determined from the dimension of the scene interaction behavior of the released service, and the second released service interest information comprises subsequent released service interest points corresponding to the scene interaction behavior big data.

The target release service may be any release service that a service registered user may pass through in the course of behavior operation, or a preset release service, and the subsequent release service interest points refer to release service interest points that have not been comprehensively determined and analyzed by the analysis manner of each dimension.

Step A130, determining the comprehensive releasing service interest data of the target releasing service according to the first releasing service interest information and the second releasing service interest information.

In order to realize the correlation consideration between the big data information with different dimensions, the first released service interest information and the second released service interest information can be integrated, the integrated released service interest data corresponding to the target released service is determined, and the released service interest integrated data comprises the subsequent released service interest points determined based on the big data information with various dimensions and the interest point attributes corresponding to the subsequent released service interest points. For example, by merging the multidimensional subsequent service delivery interest points, the data range according to which the service delivery interest points for the service registered users in the target service delivery are determined can be expanded. For example, when a service delivery interest point that a certain service registered user does not pass through on a target service delivery can not be mined through user behavior big data, although the service delivery interest point may not be embodied in a subsequent service delivery interest point corresponding to first service delivery interest information, the service delivery interest point can be crawled by a scene interaction behavior crawling program, so that the service delivery interest point is embodied in a subsequent service delivery interest point corresponding to a second service delivery interest, and finally can also be embodied in comprehensive service delivery interest data, thereby realizing more comprehensive mining and analysis of the service delivery interest point.

It can be understood that, because different manners for crawling the big data information with different dimensions are different, the subsequent service delivery interest points determined based on the big data information with different dimensions may correspond to different interest attribute information.

Therefore, the initial position information of the subsequent service delivery interest points determined based on the first service delivery interest information and the second service delivery information may correspond to different interest attribute information, so that the service delivery interest points can be accurately clustered for convenience of subsequent processing, and the subsequent service delivery interest points determined based on the big data information with different dimensions can be unified into the same information delivery service when the comprehensive service delivery interest data is determined.

Step A140, based on the service node order of the service delivery, sequentially selecting the subsequent service delivery interest points and the corresponding interest point attributes in the service window from the comprehensive service delivery interest data through the service window.

It can be understood that, when crawling different dimensions of the released service information, each crawling process is independent, so that whether other processes crawl the subsequently released service interest point or not is not considered when crawling released service information related to the subsequently released service interest point. Therefore, in the integrated service delivery interest data including the subsequent service delivery interest points determined based on the multidimensional big data information, there is a high probability that a plurality of subsequent service delivery interest points actually correspond to the same service delivery interest point in the target service delivery.

Based on this, in order to finally output a more accurate detection result of the delivered service, the subsequent service-delivered interest points in the integrated service-delivered interest data can be clustered, that is, the subsequent service-delivered interest points corresponding to the same service-delivered interest point are determined from the multiple subsequent service-delivered interest points.

It can be understood that, because the number of the subsequent service delivery interest points determined by the multidimensional big data information may be large, if all the subsequent service delivery interest points in the comprehensive service delivery interest data are clustered at the same time, the data processing amount may be too large, and more accurate clustering is difficult to perform. Based on the method, in order to cluster the subsequent service delivery interest points in the comprehensive service delivery interest data as comprehensively and accurately as possible, a service window mode can be introduced, part of the subsequent service delivery interest points in the comprehensive service delivery interest data are crawled through the service window each time for clustering, and clustering processing of all the subsequent service delivery interest points in the comprehensive service delivery interest data is realized through multi-time service window crawling and clustering.

In order to avoid missing the subsequent service delivery interest points determined from the target service delivery, the subsequent service delivery interest points in the integrated service delivery interest data may be obtained based on the service node order of the service delivery. The business window may refer to a comparison window used for summarizing a business service range. In the embodiment of the present disclosure, based on the order of service nodes for delivering services, the subsequent service-delivery interest points and the corresponding interest point attributes in the service window may be sequentially selected from the comprehensive service-delivery interest data through the service window, that is, the service-delivery interest points to be identified, whose corresponding interest point attributes are in the range corresponding to the service window, are obtained.

Step A150, clustering the subsequent service delivery interest points according to the interest point attributes selected by the service window, and determining the service delivery interest points aiming at the service registered users in the target service delivery.

And aiming at the subsequent service delivery interest points in each service window, clustering the subsequent service delivery interest points based on a plurality of clustering standards, such as attribute matching degree between the subsequent service delivery interest points, the number of the subsequent service delivery interest points and the like. Finally, the service delivery interest points determined after the clustering processing can be obtained, and the service delivery interest points are service delivery interest points aiming at service registered users in the target service delivery.

According to the technical scheme, in order to reduce the data processing pressure as much as possible and improve the confidence coefficient of the user for determining the behavioral interest of the service release, the method can acquire the user behavior big data and the scene interaction behavior big data respectively crawled by the user behavior crawling program and the scene interaction behavior crawling program of the service registered user, wherein the user behavior big data can embody the interest information from the dimensionality of the user behavior, and the scene interaction behavior big data can embody the interest information from the dimensionality of the scene interaction behavior. According to the user behavior big data and the scene interaction behavior big data, first release service interest information and second release service interest information corresponding to a target release service can be respectively determined, comprehensive release service interest data of the target release service can be determined through the first release service interest information and the second release service interest information, and the release service interest comprehensive data comprises subsequent release service interest points and corresponding interest point attributes in the target release service determined from two dimensions of user behavior and actual release service scene interaction behavior. In order to comprehensively determine the actual interest of the released service on the target released service, the subsequent interest points of the released service and the corresponding interest point attributes of the released service in the service window can be sequentially selected from the comprehensive interest data of the released service through the service window based on the sequence of the service nodes of the released service, the subsequent interest points of the released service are clustered according to the selected interest point attributes of the service window, and the interest points of the released service for the service registered user in the target released service are determined, so that the crawled multi-dimensional interest information can be comprehensively determined, and the accuracy of determining the interest points is improved.

The scene interaction behavior crawling program can comprise multiple composition modes, different composition modes can be different when the scene interaction behavior big data is crawled, and the scene interaction behavior crawling program with the multiple composition modes is developed and introduced.

In one embodiment, the scene interaction behavior crawling program may be composed of a plurality of sets of crawling programs, and may include a first crawling program and a second crawling program, for example. The first crawling program can comprise a first scene interaction behavior crawling thread and a second scene interaction behavior crawling thread, and data crawling parameters of the first scene interaction behavior crawling thread and the second scene interaction behavior crawling thread are different. Therefore, the crawling process of the two scene interaction behaviors can crawl the scene interaction behaviors from different visual angles, the linkage consideration effect on the scene interaction behavior information is realized, the finally crawled scene interaction behavior big data is more comprehensive, the reliability is higher, and the identification of the service release interest points is more accurate.

The interactive behavior data can be analyzed and processed using models based on deep learning techniques. The method comprises the steps of firstly, determining first interest sub-information according to first interaction behavior data obtained by a first scene interaction behavior crawling process, and determining second interest sub-information according to second interaction behavior data obtained by a second scene interaction behavior crawling process, wherein the first interest sub-information and the second interest sub-information respectively comprise subsequent service delivery interest points determined based on the first interaction behavior data and the second interaction behavior data.

It can be understood that when subsequent released service interest points are determined through scene interaction behaviors, due to interference of various factors, individual interest points with lower confidence degrees may exist in the determined subsequent released service interest points. Based on this, after the first interest sub-information and the second interest sub-information are determined, the incredible subsequent service delivery interest points can be screened out from the first interest sub-information and the second interest sub-information through the classification model, and then the screened first interest sub-information and the screened second interest sub-information determine service delivery interest sub-information corresponding to the first crawling program in the second service delivery interest information.

Besides the first crawling program, the scene interaction behavior crawling program can further comprise a second crawling program, and the second crawling program can be a combined scene interaction behavior crawling program formed by two dynamically coordinated scene interaction behavior crawling programs.

When the combined scene interaction behavior crawling program is used for detecting service delivery interest points, two dynamically coordinated scene interaction behavior crawling programs can be used for crawling coordinator interaction behavior data and coordinated party interaction behavior data respectively, scene interaction behavior big data can comprise the coordinator interaction behavior data and the coordinated party interaction behavior data, the coordinator interaction behavior data refers to scene interaction behaviors crawled out by the dynamically coordinated scene interaction behavior crawling program located on the left side in the combined scene interaction behavior crawling program, and the coordinated party interaction behavior data refers to scene interaction behaviors crawled out by the dynamically coordinated scene interaction behavior crawling program located on the right side in the combined scene interaction behavior crawling program.

It can be understood that, if two dynamically coordinated scene interaction behavior crawling processes crawl the same interaction behavior at the same time, in a normal case, the objects in the interaction set corresponding to the interaction behavior nodes in the coordinating party interaction behavior data and the coordinated party interaction behavior data should be consistent. Based on the rule, interactive behavior nodes corresponding to the same interactive behavior in the coordinating party interactive behavior data and the coordinated party interactive behavior data can be determined by analyzing interactive concentrated objects of the interactive behavior nodes in the coordinating party interactive behavior data and the coordinated party interactive behavior data.

Based on this, in the embodiment of the present disclosure, in order to determine the sub information of interest in launching the service corresponding to the second crawling program, the interaction centralized object of the interaction behavior node in the coordinator interaction behavior data and the coordinated party interaction behavior data may be analyzed. In an embodiment, the method may determine, through a machine learning network, release business interest sub-information corresponding to a second crawling program in the second release business interest information according to coordinator interaction behavior data and coordinated party interaction behavior data crawled by a second crawling program, where the release business interest sub-information includes subsequent release business interest points determined based on the coordinator interaction behavior data and the coordinated party interaction behavior data.

Step A160, obtaining the data related to the first product information of the service-launching interest point in the first product information push service and the data related to the second product information of the second product information push service.

In one embodiment, the service delivery interest point may refer to a service delivery unit that may be interested by a registered user for a target service delivery, such as a newly added function unit, a commodity, a news link, an application software service, and the like.

Furthermore, the first product information pushing service and the second product information pushing service are a local product information pushing service and a remote product information pushing service corresponding to the service social relationship circle where the service registered user is located. The first product information pushing service and the second product information pushing service are product information pushing services preset by the cloud server, and pushing emphasis of the first product information pushing service and pushing emphasis of the second product information pushing service can be different. For example, the first product information push service may focus on a local service corresponding to a service social relationship circle where the service registered user is located, and the second product information push service may focus on a remote service corresponding to the service social relationship circle where the service registered user is located. For another example, the first product information push service may focus on a remote service corresponding to a service social relationship circle where the service registered user is located, and the second product information push service may focus on a local service corresponding to the service social relationship circle where the service registered user is located.

Accordingly, the first product information-related data and the second product information-related data may be different, and the following description may be referred to for the first product information-related data and the second product information-related data.

Step A170, analyzing the first product information related data and the second product information related data, and determining first service matching information between the service delivery interest point and a preset information delivery plan and second service matching information between the service delivery interest point and a preset information mining plan.

In one embodiment, the information delivery plan and the information mining plan may be understood as a service plan updated based on an information delivery policy, for example, the information delivery plan is used for characterizing a process of delivering behavior data of a service delivery interest point relative to a target delivery service, and the information mining plan is used for characterizing a process of delivering content data mining of a cloud server relative to the target delivery service.

Further, the first service matching information between the service delivery interest point and the preset information delivery plan can be understood as matching information between the service delivery interest point and the information delivery plan on the service plan execution level, and the second service matching information between the service delivery interest point and the preset information mining plan can be understood as matching information between the service delivery interest point and the information mining plan on the service plan execution level. Generally speaking, the first service matching information and the second service matching information may be partially related, and the first service matching information and the second service matching information may reflect an adaptation situation between the service delivery interest point and different service plans, so as to facilitate subsequent determination of a service information delivery policy.

Step A180, determining target service matching information of the service delivery interest points based on the first service matching information, the second service matching information and the service matching information between the information delivery plan and the information mining plan.

In one embodiment, the target service matching information of the service delivery interest point includes: (1) the service plan execution condition of the service execution active party is the service delivery service interest point, (2) the service plan execution condition of the service execution passive party is the service delivery service interest point, and (3) the service plan execution condition of the service execution passive party is the service delivery service interest point. Therefore, the target service matching information of the service delivery interest point can be completely and accurately determined by analyzing the first service matching information, the second service matching information and the service matching information between the information delivery plan and the information mining plan.

Step A190, determining a service information delivery strategy for delivering the service interest points according to the target service matching information and the service interest knowledge map information.

Based on the steps, the related data of different product information of the service delivery interest points under different product information push services can be analyzed to determine first service matching information between the service delivery interest points and the information delivery plan and second service matching information between the service delivery interest points and the information mining plan, so that the first service matching information, the second service matching information and the service matching information between different service execution plans can be comprehensively analyzed to realize high matching with the current service execution plan, and further, the target service matching information can be accurately determined. Therefore, the service information delivery strategy which is as suitable as possible for the current service execution plan can be determined by combining the target service matching information and the service interest knowledge map information, so that the information pushing accuracy of the service information delivery strategy is ensured, accurate and effective delivery strategy rule updating can be carried out on the service delivery interest points through the service information delivery strategy, and user complaint behaviors caused by low reliability of the service information delivery strategy are avoided.

Fig. 3 is a schematic functional module diagram of a cloud service mining device 300 based on 5G interconnection big data according to an embodiment of the present disclosure, and the functions of the functional modules of the cloud service mining device 300 based on 5G interconnection big data are explained in detail below.

The obtaining module 310 is configured to generate corresponding cloud service delivery data according to a service information delivery policy determined based on a service delivery interest point of a service registered user of the 5G internet terminal, and obtain service feedback data of the service registered user on the cloud service delivery data.

The extracting module 320 is configured to perform feature extraction on the service feedback data according to the preset social relationship features to obtain target social influence features, where the service feedback data includes a target knowledge entity, the preset social relationship features include a plurality of social relationship influence degrees, and each social relationship influence degree is used to indicate a confidence that one social user tends to a cloud service where any knowledge entity is located in any user service data.

The updating module 330 is configured to determine mining cloud service information of the service feedback data according to the target social influence characteristic, and update the information push weight of the target delivery cloud service in the service information delivery policy based on the mining cloud service information.

Fig. 4 illustrates a hardware structure diagram of a big data pushing system 100 for implementing the above-mentioned cloud service mining method based on 5G interconnected big data according to an embodiment of the present disclosure, and as shown in fig. 4, the big data pushing system 100 may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

In a specific implementation process, at least one processor 110 executes computer-executable instructions stored in the machine-readable storage medium 120, so that the processor 110 may execute the cloud service mining method based on 5G interconnection big data according to the above method embodiment, the processor 110, the machine-readable storage medium 120, and the communication unit 140 are connected through the bus 130, and the processor 110 may be configured to control a transceiving action of the communication unit 140, so as to perform data transceiving with the aforementioned 5G internet terminal 200.

For a specific implementation process of the processor 110, reference may be made to the above-mentioned method embodiments executed by the big data pushing system 100, which implement principles and technical effects similar to each other, and details of this embodiment are not described herein again.

In addition, a readable storage medium is also provided, where a computer execution instruction is preset in the readable storage medium, and when a processor executes the computer execution instruction, the above cloud service mining method based on 5G interconnection big data is implemented.

Finally, it should be understood that the examples in this specification are only intended to illustrate the principles of the examples in this specification. Other variations are also possible within the scope of this description. Accordingly, by way of example, and not limitation, alternative configurations of the embodiments of the specification can be seen as matching the teachings of the specification. Accordingly, the embodiments of the present description are not limited to only those embodiments explicitly described and depicted herein.

Claims (10)

1. A cloud service mining method based on 5G interconnected big data is characterized by being applied to a big data pushing system, wherein the big data pushing system is in communication connection with a plurality of 5G interconnected network terminals, and the method comprises the following steps:

generating corresponding cloud service delivery data according to a service information delivery strategy determined based on the service delivery service interest points of the service registered users of the 5G internet terminal, and acquiring service feedback data of the service registered users on the cloud service delivery data;

performing feature extraction on the service feedback data according to preset social relationship features to obtain target social influence features, wherein the service feedback data comprise target knowledge entities, the preset social relationship features comprise a plurality of social relationship influence degrees, and each social relationship influence degree is used for representing the confidence degree that a social user tends to a cloud service where any knowledge entity is located in any user service data;

and determining mining cloud service information of the service feedback data according to the target social influence characteristics, and updating the information push weight of the target delivery cloud service in the service information delivery strategy based on the mining cloud service information.

2. The cloud service mining method based on 5G interconnection big data according to claim 1, wherein the step of determining the mining cloud service information of the service feedback data according to the target social influence feature comprises:

performing feature restoration on the target social influence features to obtain first mining cloud service information of the service feedback data, wherein the first mining cloud service information is used for indicating a target cloud service influenced by the target knowledge entity in the service feedback data;

according to the preset social relationship characteristics, performing user service data conversion on the first mining cloud service information to obtain a plurality of pieces of benchmark mining cloud service information, wherein each piece of benchmark mining cloud service information corresponds to one social relationship influence degree and is used for indicating the target cloud service inclined by the corresponding social user;

and processing the target social influence characteristics based on the plurality of benchmark mining cloud service information to obtain second mining cloud service information of the service feedback data.

3. The cloud service mining method based on 5G interconnection big data according to claim 2, wherein the processing the target social influence feature based on the plurality of pieces of benchmark mining cloud service information to obtain second mining cloud service information of the service feedback data comprises:

determining user service migration behavior data according to service operation behavior changes among the plurality of pieces of benchmark mining cloud service information, wherein the user service migration behavior data is used for indicating the service operation behavior changes among the target cloud services indicated by the plurality of pieces of benchmark mining cloud service information;

and fusing the target social influence characteristics with the user service migration behavior data to obtain the second mining cloud service information.

4. The cloud service mining method based on 5G interconnection big data according to claim 3, wherein each datum mining cloud service information comprises a first distribution probability corresponding to a plurality of service feedback objects in the service feedback data, and the first distribution probability is used for representing a probability value of the corresponding service feedback object in the target cloud service;

the determining user service migration behavior data according to the service operation behavior change among the plurality of pieces of benchmark mining cloud service information includes:

determining difference user service data between the plurality of benchmark mining cloud service information and election user service data of the plurality of benchmark mining cloud service information;

determining the square sum of tendency values of service feedback objects positioned at the same service node in a plurality of different user service data;

determining the evolution of the ratio of the square sum corresponding to each service node to the number of the plurality of datum mining cloud service information as a second distribution probability of each service node;

and constructing the user service migration behavior data according to the second distribution probability of the plurality of service nodes.

5. The cloud service mining method based on 5G interconnection big data according to claim 3, wherein the fusing the target social influence feature and the user service migration behavior data to obtain the second mining cloud service information comprises:

determining election user service data of the plurality of datum mining cloud service information;

determining the fusion feature of the target social influence feature and the user service migration behavior data and the weighting feature of the target social influence feature as a first fusion social influence feature;

determining the fusion feature of the target social influence feature and the election user service data and the weighting feature of the target social influence feature as a second fusion social influence feature;

splicing the first fusion social influence feature and the second fusion social influence feature to obtain a spliced fusion social influence feature;

and performing deep feature extraction processing on the splicing and fusing social influence features to obtain the second mining cloud service information.

6. The cloud service mining method based on 5G interconnection big data according to any one of claims 1 to 5, wherein the step of performing feature extraction on the service feedback data according to the preset social relationship features to obtain target social influence features comprises the steps of:

extracting the characteristics of the service feedback data to obtain a first social influence characteristic of the service feedback data;

fusing the preset social relationship characteristic with the first social influence characteristic to obtain a second social influence characteristic;

and performing feature restoration on the second social influence feature to obtain the target social influence feature.

7. The cloud service mining method based on 5G interconnection big data according to claim 1, wherein the step of performing feature extraction on the service feedback data according to the preset social relationship features to obtain the target social influence features is performed by a first cloud service mining network;

the step of performing feature restoration on the target social influence features to obtain first mining cloud service information of the service feedback data is executed by the first cloud service mining network;

the step of performing user service data conversion on the first mining cloud service information according to the preset social relationship characteristics to obtain a plurality of pieces of benchmark mining cloud service information is executed by a user service data conversion model;

the step of processing the target social influence characteristics based on the plurality of benchmark mining cloud service information to obtain second mining cloud service information of the service feedback data is executed by a second cloud service mining network;

wherein the method further comprises:

acquiring reference service feedback data, a plurality of reference datum mining cloud service information and the preset social relationship characteristics, wherein the reference service feedback data comprises reference knowledge entities, each reference datum mining cloud service information corresponds to a social relationship influence degree, and each reference datum mining cloud service information is used for indicating a reference cloud service influenced by the reference knowledge entities inclined by corresponding social users in the reference service feedback data;

calling the first cloud service mining network, and performing feature extraction on the reference service feedback data according to the preset social relationship features to obtain target reference social influence features;

calling the first cloud service mining network, and performing feature restoration on the target reference social influence feature to obtain first reference mining cloud service information of the reference service feedback data, wherein the first reference mining cloud service information is used for indicating the reference cloud service influenced by the reference knowledge entity in the reference service feedback data;

calling the user service data conversion model, and performing user service data conversion on the first reference mining cloud service information according to the preset social relationship characteristics to obtain a plurality of prediction benchmark mining cloud service information, wherein each prediction benchmark mining cloud service information corresponds to a social relationship influence degree, and each prediction benchmark mining cloud service information is used for indicating the predicted reference cloud service inclined by the corresponding social user;

calling the second cloud service mining network, and processing the target reference social influence characteristics based on the plurality of prediction benchmark mining cloud service information to obtain prediction mining cloud service information of the reference service feedback data;

according to the preset social relationship characteristics, performing weighted fusion on the plurality of reference datum mining cloud service information to obtain fusion datum mining cloud service information;

and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the service operation behavior change between the forecast mining cloud service information and the fusion benchmark mining cloud service information.

8. The cloud service mining method based on 5G interconnection big data according to claim 7, wherein the training of the first cloud service mining network, the user service data conversion model, and the second cloud service mining network according to the service operation behavior change between the prediction mining cloud service information and the fusion benchmark mining cloud service information includes:

determining a first model evaluation index value according to the service operation behavior change between the forecast excavation cloud service information and the fusion benchmark excavation cloud service information;

according to the first model evaluation index value, training the first cloud service mining network, the user service data conversion model and the second cloud service mining network;

wherein the evaluating the index value according to the first model to train the first cloud service mining network, the user service data conversion model and the second cloud service mining network includes:

determining a second model evaluation index value according to the service operation behavior change between the first reference mining cloud service information and the fusion benchmark mining cloud service information; training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the first model evaluation index value and the second model evaluation index value; or

Determining a third model evaluation index value according to the service operation behavior change between the plurality of pieces of prediction benchmark mining cloud service information and the corresponding pieces of reference benchmark mining cloud service information; training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the first model evaluation index value and the third model evaluation index value;

the user service data conversion model comprises a feature extraction unit, a feature fusion unit and a feature restoration unit;

the calling the user service data conversion model, and performing user service data conversion on the first reference mining cloud service information according to the preset social relationship characteristics to obtain a plurality of prediction benchmark mining cloud service information, includes:

splicing the reference service feedback data and the first reference mining cloud service information to obtain first reference spliced user service data;

calling the feature extraction unit to extract features of the first reference splicing user service data to obtain a first reference social influence feature;

calling the feature fusion unit, and fusing the preset social relationship feature and the first reference social influence feature to obtain a second reference social influence feature;

and calling the feature restoration unit to perform feature restoration on the second reference social influence feature to obtain the plurality of prediction benchmark mining cloud service information.

9. The cloud service mining method based on 5G interconnection big data according to claim 8, wherein the method further comprises:

splicing the reference service feedback data and the fusion benchmark mining cloud service information to obtain second reference spliced user service data;

calling the feature extraction unit to extract features of the second reference splicing user service data to obtain a third reference social influence feature;

the evaluating the index value according to the first model, and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network includes:

determining a fourth model evaluation index value according to the business operation behavior change between the third reference social influence characteristic and the first reference social influence characteristic;

and training the first cloud service mining network, the user service data conversion model and the second cloud service mining network according to the first model evaluation index value and the fourth model evaluation index value.

10. A big data pushing system, comprising a processor, a machine-readable storage medium, and a communication unit, wherein the machine-readable storage medium, the communication unit, and the processor are associated through a bus system, the communication unit is configured to be communicatively connected to at least one 5G internet terminal, the machine-readable storage medium is configured to store computer instructions, and the processor is configured to execute the computer instructions in the machine-readable storage medium to perform the cloud service mining method based on 5G internet big data according to any one of claims 1 to 9.

Images