CN112925831A

CN112925831A - Big data mining method and big data mining service system based on cloud computing service

Info

Publication number: CN112925831A
Application number: CN202110354789.0A
Authority: CN
Inventors: 王琪; 王林军
Original assignee: Individual
Current assignee: Individual
Priority date: 2021-04-01
Filing date: 2021-04-01
Publication date: 2021-06-08

Abstract

The embodiment of the disclosure provides a cloud computing service-based big data mining method and a big data mining service system, online acquisition behavior configuration data of a second cloud service process is fed back to the big data mining service system, the big data mining service system applies the online acquisition behavior configuration data to an online acquisition behavior strategy in the second cloud service process through a big data registration terminal, so that online acquisition jump execution information of a first cloud service process is converted into online acquisition jump execution information of the second cloud service process, and under the combined action of the online acquisition jump execution information of the second cloud service process and the online acquisition migration execution information of the second cloud service process, online knowledge map information of the first cloud service process is converted into online knowledge map information of the second cloud service process, so that the optimization of online acquisition objects in an online acquisition behavior strategy is automatically controlled, and the accuracy of big data acquisition is improved.

Description

Big data mining method and big data mining service system based on cloud computing service

Technical Field

The disclosure relates to the technical field of big data acquisition, in particular to a big data mining method and a big data mining service system based on cloud computing service.

Background

Big data acquisition refers to the process of acquiring data from sensors and intelligent devices, enterprise online systems, enterprise offline systems, social networks, internet platforms, and the like. The data comprises various types of structured, semi-structured and unstructured mass data such as RFID data, sensor data, user behavior data, social network interaction data, mobile internet data and the like. The data acquisition method has the advantages of various data sources, various data types, large data quantity and high generation speed, and the traditional data acquisition method is completely insufficient. Therefore, the big data collection technology faces many technical challenges, and on one hand, the reliability and the efficiency of data collection need to be ensured, and meanwhile, repeated data also need to be avoided. In the related art, how to improve the accuracy of the key big data acquisition is a technical problem to be considered and solved urgently.

Disclosure of Invention

In order to overcome at least the above disadvantages in the prior art, an object of the present disclosure is to provide a big data mining method and a big data mining service system based on cloud computing service.

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

acquiring online knowledge map information of an online acquisition object of an online acquisition tool of the big data registration terminal in a first cloud service flow in an online acquisition behavior strategy, online acquisition skip execution information of the online acquisition behavior strategy in the first cloud service flow, and online acquisition migration execution information of the online acquisition behavior strategy in the first cloud service flow;

training a target machine learning network based on online knowledge map information of the first cloud service flow, online acquisition skip execution information of the first cloud service flow and online acquisition migration execution information of the first cloud service flow to obtain online acquisition behavior configuration data for configuring an online acquisition behavior strategy in a second cloud service flow, wherein the second cloud service flow is in the migration cloud service flow of the first cloud service flow;

and applying the online acquisition behavior configuration data to the online acquisition behavior strategy in the second cloud service process so as to conveniently perform big data mining on the acquired key big data of the big data operation activity after performing the key big data acquisition of the big data operation activity based on the online acquisition behavior strategy.

In a second aspect, the disclosed embodiment further provides a cloud computing service-based big data mining system, where the cloud computing service-based big data mining system includes a big data mining service system and a plurality of big data registration terminals communicatively connected to the big data mining service system;

the big data mining service system is used for:

According to any one of the above aspects, the present disclosure feeds back the online collection behavior configuration data of the second cloud service process to the big data mining service system, and the big data mining service system applies the online collection behavior configuration data to the online collection behavior policy in the second cloud service process through the big data registration terminal, thereby converting the on-line acquisition skip execution information of the first cloud service process to the on-line acquisition skip execution information of the second cloud service process, and under the combined action of the online acquisition skip execution information of the second cloud service flow and the online acquisition migration execution information of the second cloud service flow, converting the online knowledge graph information of the first cloud service flow into the online knowledge graph information of the second cloud service flow, thereby realizing the optimization of an online acquisition object in an automatic control online acquisition behavior strategy.

Drawings

Fig. 1 is a schematic view of an application scenario of a cloud computing service-based big data mining system according to an embodiment of the present disclosure;

fig. 2 is a schematic flowchart of a cloud computing service-based big data mining method according to an embodiment of the present disclosure;

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

fig. 4 is a schematic block diagram of structural components of a big data mining service system for implementing the above big data mining method based on cloud computing services according to the embodiment of the present disclosure.

Detailed Description

The present disclosure is described in detail below with reference to the drawings, and the specific operation methods in the method embodiments can also be applied to the device embodiments or the system embodiments.

Fig. 1 is an interaction diagram of a cloud computing service-based big data mining system 10 provided by an embodiment of the present disclosure. The cloud computing service-based big data mining system 10 may include a big data mining service system 100 and a big data registration terminal 200 communicatively connected to the big data mining service system 100. The cloud computing service-based big data mining system 10 shown in fig. 1 is only one possible example, and in other possible embodiments, the cloud computing service-based big data mining system 10 may also include only at least some of the components shown in fig. 1 or may also include other components.

In a separate embodiment, the big data mining service system 100 and the big data registration terminal 200 in the cloud computing service-based big data mining system 10 may cooperatively perform the cloud computing service-based big data mining method described in the following method embodiments, and the detailed description of the method embodiments may be referred to in the following steps of the big data mining service system 100 and the big data registration terminal 200.

Step S110, obtaining an activity trigger event set corresponding to each big data operation activity according to the online collection behavior policy of the online collection tool of the big data registration terminal 200.

The activity triggering event set can comprise activity triggering event data of at least two target business data sources, and the activity triggering event data comprises an activity triggering event and a request digital object when the target business data sources request the target cloud computing business resources each time. The business data source may refer to a data source where a data providing or presentation service exists, and the activity triggering event may refer to a triggering behavior of a user for data reading in a page of the business data source. Requesting a digital object may refer to a digital service specifically referred to, but is not limited to.

In a separate embodiment, step S110 may be implemented by the following steps.

Step S111, obtaining the activity triggering event data of each target service data source. The activity triggering event data comprises a big data operation activity signature, an activity triggering event and a request digital object when the target business data source requests the target cloud computing business resources each time, and the big data operation activity signature is used for indicating the target business data source to obtain the big data operation activity of the target cloud computing business resources.

In this embodiment, the current key cloud computing service resource may be used as a target cloud computing service resource, a service data source configured with the target cloud computing service resource may be used as a target service data source, and an activity triggering event crawling script is added in a configuration packet of the target cloud computing service resource, so that each time the target service data source requests the target cloud computing service resource, the activity triggering event crawling script can acquire an activity triggering event and a request digital object of the current target service data source, report the acquired activity triggering event and the request digital object to the big data mining service system, and report a data source signature and a big data operation activity signature of the target service data source to the big data mining service system, where the data source signature is used to uniquely identify the target service data source, and the big data operation activity signature is used to indicate a source of the target cloud computing service resource in the target service data source, after receiving information reported by an activity triggering event crawling script, a big data mining service system summarizes an activity triggering event, a request digital object, a big data operation activity signature and a data source signature belonging to the same target business data source under a corresponding target business data source name according to the data source signature to obtain activity triggering event data of each target business data source, wherein the activity triggering event data can be regarded as a triggering sequence, and each data in the triggering sequence corresponds to a request action of the target business data source on a target cloud computing business resource. For example, the target cloud computing service resource requests n (n ≧ 1 integer) times on the target service data source a, the activity trigger event data of the target service data source a includes n pieces of data.

Step S112, clustering the activity triggering event data of each target business data source according to a big data operation activity signature to obtain reference activity triggering event clusters corresponding to the big data operation activity indicated by each big data operation activity signature, wherein each reference activity triggering event cluster comprises the activity triggering event data of at least two target business data sources, and all the target business data sources in the same reference activity triggering event cluster have the same big data operation activity signature.

After the activity triggering event data of each target business data source is obtained, the activity triggering event data can be clustered according to the big data operation activity signature of the activity triggering event data, namely, the activity triggering event data with the same big data operation activity signature are classified into the same reference activity triggering event cluster, so that the big data operation activity signature of each reference activity triggering event cluster is different, and different reference activity triggering event clusters correspond to different big data operation activities because the big data operation activity signature indicates the source big data operation activities of the target cloud computing business resources. The step classifies the activity triggering event data of the target business data source according to the big data operation activities, and facilitates the follow-up analysis of the detail characteristics of each big data operation activity based on the activity triggering event characteristics of the target business data source.

Since if too few trigger spans exist in a certain big data operation activity, the first big data operation information may have a large error, and therefore, the big data operation activity with too few trigger spans needs to be eliminated. Therefore, after step S112, step S113 may also be included: and counting the trigger span of the activity trigger event data in each reference activity trigger event cluster, and taking the reference activity trigger event cluster with the trigger span larger than or equal to a preset threshold value as an activity trigger event set. By removing the reference activity triggering event cluster with the triggering removal span smaller than the preset threshold value, the activity triggering event set for analyzing the importance of the big data operation activity is obtained, and the accuracy of the calculation of the subsequent first big data operation information can be improved.

Step S120, determining first big data operation information of each activity trigger event set according to the activity trigger event when each target business data source requests the target cloud computing business resources, wherein the first big data operation information represents the big data intention degree of the target business data sources in the activity trigger event set when the target cloud computing business resources are requested.

Wherein the first big data operation information of each activity triggering event set comprises first member big data operation information and second member big data operation information. In an independent embodiment, step S120 may specifically include: determining first member big data operation information of each activity triggering event set according to activity triggering events when each target service data source requests a target cloud computing service resource for the first time; the first member big data operation information represents the big data intention degree of a target business data source in the activity triggering event set when applying for a target cloud computing business resource; determining second member big data operation information of each activity trigger event set according to activity trigger events when each target business data source requests the target cloud computing business resources for the first time in each big data operation activity; the second member big data operation information represents the big data intention degree of a target business data source in the activity triggering event set when the target cloud computing business resources are started; and taking the first member big data operation information and the second member big data operation information of each activity triggering event set as the first big data operation information of the activity triggering event set.

In a separate embodiment, the following steps may be included in response to big data intent detection for a big data operation activity based on a plurality of activity triggering events.

Step S301, taking an activity trigger event corresponding to when each target service data source first requests a target cloud computing service resource in each big data operation activity as a second activity trigger event.

Step S302, determining the target service data source with the event amount of the second activity triggering event being greater than or equal to the target event amount as the reference service data source.

The large data intention object is expected to be mined by detecting the change conditions of a plurality of activity trigger events corresponding to the target cloud computing service resources before and after the target service data source logs in, so that the data elimination does not only eliminate the activity trigger event in the first configuration, but first activity trigger event data in each period are eliminated from the same target service data source periodically as second activity trigger events, and then the second activity trigger event data of the same target service data source are aggregated. That is, one or more second activity triggering events can be obtained for the same target business data source. And removing all second activity triggering events of the same aggregated target business data source, namely removing the target business data source of which the triggering span of the second activity triggering event is smaller than the target event quantity n, taking the target business data source of which the event quantity of the second activity triggering event is greater than or equal to n as a reference business data source, wherein n is a defined threshold value and can be defined as 5-10, and the purpose is to reduce the influence of the characteristics of the information of which the second activity triggering event quantity is too small on the overall result.

Step S303, performing feature processing on all the second activity trigger events of each reference service data source to obtain second activity trigger event features corresponding to each reference service data source.

After all the second activity trigger events of each reference service data source are obtained, feature processing needs to be performed on multiple second activity trigger events of each reference service data source, and second activity trigger event features corresponding to the reference service data sources are generated. Wherein the performing feature processing on the plurality of second activity triggering events of the reference service data source comprises: features for big data intent analysis are extracted from the second activity trigger event and the extracted features are processed into feature vectors that are available for computation. Since this part detects the change of a plurality of activity triggering events, the adopted features are basically dynamic features, namely the features which change along with the user behavior. Further, the extracted features are processed into feature vectors which can be used for calculation, and mainly, changes of the features in a preset time are converted into characters or numerical values.

Step S304, inputting the second activity triggering event characteristics of each reference business data source into a big data intention classification model, and carrying out big data intention analysis on the second activity triggering event characteristics of each reference business data source based on the big data intention classification model to obtain a second big data intention degree of each reference business data source.

After the second activity triggering event characteristics of the reference business data sources in each activity triggering event set are obtained, the target business data sources with big data intentions can be detected by adopting a big data intention classification model. In the embodiment of the disclosure, a decision tree algorithm is adopted as a big data intention classification model. For the decision tree algorithm, please refer to the above description, which is not repeated herein.

Inputting the second activity triggering event characteristics of each reference business data source into the decision tree model, and performing big data intention characteristic analysis on the second activity triggering event characteristics of each reference business data source based on the decision tree model to obtain a model big data intention score, namely a second big data intention degree, of each reference business data source.

Step S305, determining second member big data operation information corresponding to each activity trigger event set based on the total data sources of the reference business data sources in each activity trigger event set and the second big data intention degree of the reference business data sources with the second big data intention degree larger than a second threshold value.

After obtaining the second big data intention degree of each reference service data source, the reference service data sources with the second big data intention degree larger than the first threshold (which may be 0.8) may be selected, and these reference service data sources may be regarded as the pending big data intention objects. And calculating the total score of all the undetermined big data intention objects in each activity trigger event set according to the second big data intention degree of the undetermined big data intention objects, then dividing the total score by the total amount of the data sources of the reference business data sources in the activity trigger event set to obtain a big data intention split of each activity trigger event set, taking the big data intention split as second member big data operation information, wherein the higher the second member big data operation information is, the worse the importance of the big data operation activity is.

Step S130, determining second big data operation information of each activity triggering event set according to the request digital object when each target service data source requests the target cloud computing service resource each time; the second big data operation information represents a request service range of the target service data source in the activity triggering event set for requesting the target cloud computing service resource.

In a separate embodiment, the calculation method of the second big data operation information may include the following steps.

Step S401, according to the request digital object when each target service data source requests the target cloud computing service resource each time, counting the data source quantity of the target service data source of each activity trigger event concentrated in the first request digital object to configure the target cloud computing service resource and enable the target cloud computing service resource at the second request digital object, the data source quantity of the target service data source to configure the target cloud computing service resource at the first request digital object, the data source quantity of the target service data source to enable the target cloud computing service resource at both the first request digital object and the second request digital object, and the data source quantity of the target service data source to enable the target cloud computing service resource at the first request digital object; the second requested digital object is a requested digital object subsequent to the first requested digital object.

Step S402, according to the data source quantity of the target service data source of the target cloud computing service resource configured in the first request digital object in each activity trigger event set and the data source quantity of the target service data source of the target cloud computing service resource enabled in the second request digital object and the data source quantity of the target service data source of the target cloud computing service resource configured in the first request digital object, third member big data operation information corresponding to each activity trigger event set is calculated.

Step S403, calculating fourth member big data operation information corresponding to each activity trigger event set according to the data source amount of the target service data source, in which the first request digital object and the second request digital object both use the target cloud computing service resource, and the data source amount of the target service data source, in which the first request digital object uses the target cloud computing service resource.

Step S404, determining second big data operation information corresponding to each activity trigger event set according to the third member big data operation information and the fourth member big data operation information corresponding to the activity trigger event set.

The embodiment of the present disclosure mainly calculates two kinds of second big data operation information, which are the second big data operation information of the original activity triggering stage and the second big data operation information of the progress activity triggering stage. In the above, the target business data source configuring the target cloud computing business resource at the first request digital object is regarded as the original activity triggering phase, and the target business data source configured before the first request digital object and enabling the target cloud computing business resource at the first request digital object is regarded as the advanced activity triggering phase. The second big data operation information, that is, the third member big data operation information in the original activity triggering stage refers to a ratio of a data source amount of a target service data source, in which the activity triggering event is concentrated on the first request digital object to configure the target cloud computing service resource, and the target cloud computing service resource is enabled in the second request digital object, to a data source amount of the target service data source, in which the target cloud computing service resource is configured in the first request digital object. The second big data operation information, that is, the fourth member big data operation information in the progress activity triggering stage refers to a ratio of a data source amount of a target service data source in which the first request digital object and the second request digital object both enable the target cloud computing service resource to a data source amount of a target service data source in which the first request digital object enables the target cloud computing service resource.

The following are exemplary: the second big data operation information of the original activity triggering stage may refer to a ratio of the number of resources that the target cloud computing service resource is configured and still started at the nth node to the number of resources that the target cloud computing service resource is configured in the current day.

The second big data operation information of the progress activity triggering stage may refer to a ratio of the number of resources in which the target cloud computing service resource is started and is still started at the nth node to the number of resources in which the target cloud computing service resource is currently started.

And for each activity triggering event set, calculating a mean vector sequence of the third member big data operation information and the fourth member big data operation information of each activity triggering event set, and taking the mean vector sequence as the second big data operation information corresponding to the activity triggering event set.

Step S140, according to the first big data operation information and the second big data operation information of the activity trigger event set corresponding to each big data operation activity, key big data collection of the big data operation activity is carried out.

The first big data operation information of the activity triggering event set comprises: the method comprises the steps of obtaining first member big data operation information based on single activity triggering event big data intention detection and obtaining second member big data operation information based on a plurality of activity triggering event big data intention detection. In one embodiment, performing the key big data collection of the big data operation activity according to the first big data operation information and the second big data operation information of the activity trigger event set corresponding to each big data operation activity may include the following steps

Step S141, performing standardized conversion processing on the first member big data operation information, the second member big data operation information, and the second big data operation information of each activity trigger event set, to obtain a first standardized value corresponding to the first member big data operation information, a second standardized value corresponding to the second member big data operation information, and a third standardized value corresponding to the second big data operation information.

Step S142, obtaining a preset first member importance coefficient corresponding to the first member big data operation information, a preset second member importance coefficient corresponding to the second member big data operation information, and a preset third member importance coefficient corresponding to the second big data operation information.

And step S143, performing weight calculation on the first normalized value, the second normalized value and the third normalized value of each activity triggering event set according to the first member importance coefficient, the second member importance coefficient and the third member importance coefficient to obtain the big data intention degree of each activity triggering event set.

And step S144, acquiring key big data of the big data operation activities according to the big data intention degree of the activity trigger event set corresponding to each big data operation activity.

In this embodiment, the numerical difference between the first member big data operation information, the second member big data operation information, and the second big data operation information may be very large, and in order to combine the scores of the big data operation activities obtained by the three methods to perform big data operation activity order sorting, it is necessary to perform standardized conversion processing on the scores obtained by the three methods, where the standardized conversion processing refers to processing the first member big data operation information, the second member big data operation information, and the second big data operation information in proportion so that the first member big data operation information, the second member big data operation information, and the second big data operation information fall into a specific section. Common standardized conversion processing methods are: min-max normalization (Min-max normalization), log function transformation, atan function transformation, z-score normalization (zero-normalization), fuzzy quantization.

In one embodiment, the first big data operation information, the second big data operation information and the second big data operation information may be normalized and converted by using a normalization formula x '— (x- μ)/σ, so as to obtain a first normalized value corresponding to the first big data operation information, a second normalized value corresponding to the second big data operation information and a third normalized value corresponding to the second big data operation information, where x is a score to be processed, and is the first big data operation information, the second big data operation information or the second big data operation information, x' is a normalized value corresponding to x, μ is an average value of scores corresponding to each activity trigger event set, σ is a standard deviation of scores corresponding to each activity trigger event set, and when x is the first big data operation information, μ is an average value of the first big data operation information corresponding to each activity trigger event set, and sigma is the standard deviation of the first member big data operation information corresponding to each activity triggering event set.

Assuming that after the first member big data operation information m1, the second member big data operation information m2 and the second big data operation information m3 are subjected to the normalized conversion processing, a first normalized value m1 'of the first member big data operation information m1, a second normalized value m 2' of the second member big data operation information m2 and a third normalized value m3 ', corresponding to the second big data operation information m3 are obtained, wherein a, b and c respectively represent member importance coefficients (a, b and c are all greater than 0 and less than 1) of scores of the first member big data operation information, the second member big data operation information and the second big data operation information, the big data intention degree m of the activity trigger event set is a m 1' + b m2 '-c m 3', so that the big data operation activities corresponding to the activity trigger event set can be sequentially sorted according to the big data intention degree m, and obtaining the total sequential arrangement of the big data operation activities corresponding to the activity trigger event set, wherein the higher the score of m is, the lower the sequential arrangement is, and the higher the possibility that the corresponding big data operation activity is that the big data intends to be the big data operation activity is. Therefore, key big data collection of big data operation activities can be carried out according to the big data intention degree of the activity trigger event set corresponding to each big data operation activity. The method for collecting the key big data such as carrying out big data operation activities can be as follows: the big data operation activities corresponding to the preset number of activity trigger event sets with high big data intention degree comprise operation activity objects as key big data acquisition basis for mining, and the preset number can be a fixed value and can also be obtained by calculation according to the preset big data intention ratio and the total data source amount of all the activity trigger event sets; or calculating a difference value between the big data intention degree of each activity trigger event set and the big data intention degree with the highest score, and when the difference value is smaller than a preset difference value threshold value, taking the big data operation activity including the operation activity object corresponding to the activity trigger event set as a key big data acquisition basis for mining.

In addition, after the total sequence of the big data operation activities is sorted, the operation activity objects including the activity trigger event set with the highest sequence sorting can be used as a high-reliability key big data acquisition basis, the operation activity objects including the activity trigger event set with the lower sequence sorting can be used as a low-reliability key big data acquisition basis, and the analysis can be performed on the big data operation activities with the lower sequence sorting, for example, the difference of the starting numbers of the target cloud computing service resources in each time period in one day and the trend change difference of the line graph in the comparison analysis of the high-reliability key big data acquisition basis and the low-reliability key big data acquisition basis can cause a great problem if the starting numbers of the target cloud computing service resources in each time period in one day in the low-reliability key big data acquisition basis are basically balanced. And the difference of the starting numbers of the target cloud computing service resources on each day in a week can also be analyzed.

In the embodiment of the disclosure, an activity triggering event set corresponding to each big data operation activity is obtained, and each activity triggering event set comprises activity triggering event data of at least two target service data sources; determining first member big data operation information of each activity triggering event set according to activity triggering events when each target service data source requests a target cloud computing service resource for the first time; determining second member big data operation information of each activity trigger event set according to activity trigger events when each target business data source requests the target cloud computing business resources for the first time in each big data operation activity; determining second big data operation information of each activity triggering event set according to the request digital object when each target business data source requests the target cloud computing business resources each time; and then, according to the first member big data operation information, the second member big data operation information and the second big data operation information of the activity trigger event set corresponding to each big data operation activity, carrying out key big data acquisition on the big data operation activity. The method comprises three big data operation activity importance evaluation methods, namely first member big data operation information based on single activity trigger event big data intention detection, second member big data operation information based on multiple activity trigger event big data intention detection, and second big data operation information based on request digital objects and frequency rules reported by activity trigger events, wherein the first member big data operation information and the second member big data operation information are used for analyzing big data operation activities based on big data intention characteristics of the activity trigger events, the second big data operation information is used for analyzing big data operation activities through request service information of users, and the two methods can simultaneously combine big data intention characteristics and request service information of the activity trigger events to improve the accuracy of key big data acquisition.

According to the big data acquisition big data operation activity ordering method and device, the importance of the big data acquisition big data operation activity of the target cloud computing service resource is ordered based on the activity trigger event, and the final big data operation activity ordering result considers the order arrangement results of three big data operation activities, namely a big data intention detection result based on a single activity trigger event, a big data intention detection result based on a plurality of activity trigger events and second big data operation information. The method can output the sequence arrangement of the big data operation activities, can output the weighted summary scores of all the big data operation activities, and assists in further judging the importance of the big data operation activities. For the big data acquisition big data operation activities with low reliability, whether the weight or the frequency of the big data acquisition operation activities with low reliability is reduced or stopped can be determined after secondary analysis, and meanwhile, the weight or the frequency of the big data acquisition optimization configuration can be properly increased for the big data acquisition big data operation activities with high reliability.

It is noted that while for simplicity of explanation, the foregoing method embodiments have been described as a series of acts or combination of acts, it will be appreciated by those skilled in the art that the present disclosure is not limited by the order of acts, as some steps may, in accordance with the present disclosure, occur in other orders and concurrently. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required for the disclosure.

On the basis of the above description, some terms involved in the following embodiments of the present disclosure are explained below.

Acquiring behavior configuration data on line: and configuration project factors for influencing the big data acquisition behavior of the online acquisition object, such as big data acquisition frequency setting, big data acquisition hierarchy setting, big data acquisition interval period setting, big data acquisition continuous range setting and the like.

In the following embodiments, the online collection and migration execution information of the first cloud service process may be the real-time online collection and migration information collection acquired by the third application, that is, the online collection and migration execution information of the first cloud service process is the real online collection and migration state of the first cloud service process, or may be the previous synchronous online collection and migration data, that is, the online collection and migration execution information of the first cloud service process is the online collection and migration state simulated by the first cloud service process. The first cloud service process is a reference cloud service process relative to the second cloud service process, and units of the first cloud service process and the second cloud service process can be flexibly set according to actual development and a business iteration cycle. The online acquisition jump execution information is obtained by sensing a big data acquisition behavior in an online acquisition behavior strategy through a service script, and the online knowledge graph information is obtained by generating a knowledge graph of an online acquisition object in the online acquisition behavior strategy through the service script.

Before the foregoing step S110, as shown in fig. 2, fig. 2 is a flowchart illustrating a cloud computing service-based big data mining method provided in this embodiment of the present disclosure, and the cloud computing service-based big data mining method provided in this embodiment may be executed by the big data mining service system 100 shown in fig. 1, which is described in detail below.

Step S101, acquiring online knowledge map information of an online acquisition object in a first cloud service flow in an online acquisition behavior strategy, online acquisition skip execution information of the online acquisition behavior strategy in the first cloud service flow, and online acquisition migration execution information of the online acquisition behavior strategy in the first cloud service flow.

For example, the big data mining service system 100 may obtain online acquisition skip execution information and online knowledge graph information of a reference cloud service flow (first cloud service flow) relative to a current cloud service flow (second cloud service flow) through a service script in an online acquisition behavior policy, obtain online acquisition migration execution information of the first cloud service flow and online acquisition migration execution information of the second cloud service flow through a third application, automatically generate a configuration instruction for the online acquisition behavior policy based on the online knowledge graph information of the first cloud service flow, the online acquisition skip execution information of the first cloud service flow and the online acquisition migration execution information of the first cloud service flow, and obtain the online knowledge graph information and online knowledge graph information of the first cloud service flow, the online knowledge graph information and the online knowledge graph information of the first cloud service flow by analyzing the configuration instruction for the online acquisition behavior policy, The online acquisition skipping execution information of the first cloud service process and the online acquisition migration execution information of the first cloud service process are used for training a target machine learning network to dynamically optimize online acquisition behavior configuration data on the basis of online knowledge map information of the first cloud service process, online acquisition skipping execution information of the first cloud service process and online acquisition migration execution information of the first cloud service process in the following process.

Step S102, training a target machine learning network based on online knowledge map information of the first cloud service flow, online acquisition jump execution information of the first cloud service flow and online acquisition migration execution information of the first cloud service flow to obtain online acquisition behavior configuration data used for configuring an online acquisition behavior strategy in a second cloud service flow, wherein the second cloud service flow is the migration cloud service flow of the first cloud service flow.

For example, online knowledge map information of the first cloud service process, online acquisition skip execution information of the first cloud service process, and online acquisition migration execution information of the first cloud service process are input into the target machine learning network, so that the target machine learning network learns the following strategies: the online acquisition behavior configuration data of the online acquisition behavior strategy and the online acquisition migration execution information share online acquisition jump execution information which is used for the online acquisition behavior strategy, and the online acquisition jump execution information of the online acquisition behavior strategy is used for online knowledge map information of an online acquisition object, so that prediction processing is carried out through a target machine learning network, online acquisition behavior configuration data which are used for configuring the online acquisition behavior strategy in a second cloud service flow are obtained, namely online acquisition behavior configuration data which are used for configuring the online acquisition behavior strategy in a next cloud service flow are obtained, the online acquisition behavior of the online acquisition object in the online acquisition behavior strategy is dynamically adjusted and controlled in combination with the online acquisition migration execution information, and intelligent and accurate prediction of the online acquisition object large data acquisition behavior is further achieved.

Illustratively, in a separate embodiment, step S102 may be implemented by steps S1021-S1023: performing the following processing based on the target machine learning network: in step S1021, based on the online knowledge graph information of the first cloud service process, determining preset online knowledge graph information that meets the requirement of big data acquisition optimization in the second cloud service process; in step S1022, it is determined that the online acquisition skip execution information of the second cloud service process when the preset online knowledge graph information is realized; in step S1023, based on the linkage relationship that the online collection migration execution information of the second cloud service flow and the online collection behavior configuration data for configuring the online collection behavior policy in the second cloud service flow jointly act on the online collection skip execution information of the second cloud service flow, the online collection behavior configuration data for configuring the online collection behavior policy in the second cloud service flow is determined.

For example, the big data mining service system 100 obtains the online knowledge map information of the first cloud service process, the online collection skip execution information of the first cloud service process, and the online collection migration execution information of the first cloud service process by analyzing the configuration command for the online collection behavior policy, and training a target machine learning network to perform prediction processing, based on the strategy learned by the target machine learning network, firstly performing prediction processing on the online knowledge graph information based on the online knowledge graph information of the first cloud service process to obtain preset online knowledge graph information meeting the requirement of big data acquisition and optimization in the second cloud service process, the big data acquisition optimization requirements may be set in advance according to an actual scene, for example, the big data acquisition optimization requirements may be service conversion rate maximization of an online acquisition object, and may also be service net worth maximization brought by the online acquisition object. After the preset online knowledge map information of the second cloud service process is obtained, online acquisition jump execution information of the second cloud service process when the preset online knowledge map information is realized is determined, and finally, the online acquisition migration execution information of the second cloud service process learned based on the target machine learning network and online acquisition behavior configuration data used for configuring an online acquisition behavior strategy in the second cloud service process are jointly acted on a linkage relation of the online acquisition jump execution information of the second cloud service process, and online acquisition behavior configuration data used for configuring the online acquisition behavior strategy in the second cloud service process are determined, so that large data acquisition behaviors of online acquisition objects in the online acquisition behavior strategy are intelligently controlled.

In a separate embodiment, the online knowledge graph information of the first cloud service process is 50% of online acquisition proportion of the online acquisition object, the online knowledge graph information of the first cloud service process is subjected to prediction processing based on the online knowledge graph information of the first cloud service process, the preset online knowledge graph information of the second cloud service process is 75% of online acquisition proportion, in order to realize the preset online knowledge graph information of the second cloud service process, the online acquisition jump execution information of the second cloud service process is that the online acquisition behavior strategy is that the large data acquisition frequency reaches 8 times/second, the large data acquisition hierarchy reaches 5 span levels, and the online acquisition migration execution information of the second cloud service process is acquired through online acquisition migration information acquisition of a third party application, the global large data acquisition frequency is 5 times/second, the online collection migration execution information based on the second cloud service flow and online collection behavior configuration data used for configuring an online collection behavior strategy in the second cloud service flow jointly act on a linkage relation of online collection jump execution information of the second cloud service flow, and it is determined that the online collection behavior configuration data of the second cloud service flow is large data collection frequency set to be started to reach 8 times/second, large data collection levels are set to be started to reach 5 span levels, and a large data collection interval period is set to be 1 hour.

In a separate embodiment, a target machine learning network includes a first machine learning structure, a second machine learning structure, and a feature classification structure; performing feature association processing on the online knowledge graph information of the first cloud service process based on the first machine learning structure to obtain preset online knowledge graph information meeting big data acquisition optimization requirements in the second cloud service process; correspondingly, the feature association processing is carried out on the preset online knowledge graph information of the second cloud service process based on the graph association entity between the online acquisition jump execution information of the online acquisition behavior strategy included in the second machine learning structure and the online knowledge graph information of the online acquisition object, so that the online acquisition jump execution information of the second cloud service process is obtained when the preset online knowledge graph information of the second cloud service process is realized; on the contrary, based on the map association entity between the online acquisition behavior configuration data and the online acquisition migration execution information of the online acquisition behavior strategy included in the feature classification structure and the online acquisition jump execution information of the online acquisition behavior strategy, the online acquisition jump execution information of the second cloud service flow is mapped to the online acquisition behavior configuration data used for configuring the online acquisition behavior strategy in the second cloud service flow.

In an independent embodiment, the online knowledge graph information of the first cloud service process can be input into a first machine learning structure in a target machine learning network, the online knowledge graph information of the first cloud service process is subjected to characteristic association processing through the first machine learning structure to obtain preset online knowledge graph information meeting the requirement of big data acquisition optimization in the second cloud service process, the preset online knowledge graph information is input into the second machine learning structure, a graph association entity between online acquisition jump execution information of an online acquisition behavior strategy learned through the second machine learning structure and online knowledge graph information of an online acquisition object is subjected to characteristic association processing on the preset online knowledge graph information of the second cloud service process to obtain online acquisition jump execution information of the second cloud service process when the preset online knowledge graph information of the second cloud service process is realized, and inputting online acquisition skip execution information of the second cloud service flow into the feature classification structure, mapping online acquisition skip execution information of the second cloud service flow into online acquisition behavior configuration data used for configuring the online acquisition behavior strategy in the second cloud service flow through a map association entity between online acquisition behavior configuration data and online acquisition migration execution information of the online acquisition behavior strategy learned through the feature classification structure and online acquisition skip execution information of the online acquisition behavior strategy. Therefore, the online acquisition behavior configuration data of the second cloud service process can be predicted through the multilayer units in the target machine learning network, intelligent and accurate prediction of the online acquisition object big data acquisition behavior is achieved, and adjustment of the online acquisition object big data acquisition behavior in the online acquisition behavior strategy through a manual regulation mode is avoided.

In a separate embodiment, a target machine learning network includes a first machine learning structure, a second machine learning structure, and a feature classification structure; performing feature association processing on the online knowledge graph information of the first cloud service process based on the first machine learning structure to obtain preset online knowledge graph information meeting big data acquisition optimization requirements in the second cloud service process; on the contrary, based on a map association entity between online acquisition jump execution information of an online acquisition behavior strategy included in the second machine learning structure and online knowledge map information change of an online acquisition object, performing feature conversion processing on loss functions of online knowledge map information of the first cloud service process and preset online knowledge map information of the second cloud service process to obtain online acquisition jump execution information of the second cloud service process when the preset online knowledge map information of the second cloud service process is realized; on the contrary, based on the map association entity between the online acquisition behavior configuration data and the online acquisition migration execution information of the online acquisition behavior strategy included in the feature classification structure and the online acquisition jump execution information of the online acquisition behavior strategy, the online acquisition jump execution information of the second cloud service flow is mapped to the online acquisition behavior configuration data used for configuring the online acquisition behavior strategy in the second cloud service flow.

For example, online knowledge graph information of a first cloud service process can be input into a first machine learning structure in a target machine learning network, the online knowledge graph information of the first cloud service process is subjected to feature association processing through the first machine learning structure to obtain preset online knowledge graph information meeting big data acquisition optimization requirements in a second cloud service process, the preset online knowledge graph information is input into the second machine learning structure, a graph association entity between online acquisition skip execution information of an online acquisition behavior strategy learned through the second machine learning structure and online knowledge graph information change of an online acquisition object is obtained, feature conversion processing is performed on loss functions of the online knowledge graph information of the first cloud service process and the preset online knowledge graph information of the second cloud service process to obtain online acquisition cloud service process information of the second cloud service process when the preset online knowledge graph information of the second cloud service process is realized And the second machine learning structure inputs the online acquisition skip execution information of the second cloud service flow into the feature classification structure, and maps the online acquisition skip execution information of the second cloud service flow into online acquisition behavior configuration data used for configuring the online acquisition behavior strategy in the second cloud service flow through a map association entity between the online acquisition behavior configuration data and the online acquisition migration execution information of the online acquisition behavior strategy, which are learned through the feature classification structure, and the online acquisition skip execution information of the online acquisition behavior strategy. Therefore, the online acquisition behavior configuration data of the second cloud service process can be predicted through the multilayer units in the target machine learning network, intelligent and accurate prediction of the online acquisition object big data acquisition behavior is achieved, and adjustment of the online acquisition object big data acquisition behavior in the online acquisition behavior strategy through a manual regulation mode is avoided.

In an independent embodiment, the online knowledge graph information of the first cloud service process is that the online acquisition proportion of the online acquisition object is 50%, the preset online knowledge graph information of the second cloud service process is that the online acquisition proportion is 75%, the online knowledge graph information of the online acquisition object is changed into the online acquisition proportion increased by 25%, in order to realize the online knowledge graph information change of the online acquisition object, the online acquisition jump execution information of the second cloud service process is that the online acquisition behavior strategy has a large data acquisition frequency of 8 times/second and a large data acquisition level of 5 span levels, so that the relationship between the online acquisition jump execution information of the online acquisition behavior strategy and the online knowledge graph information change of the online acquisition object in the actual growth process of the online acquisition object is learned through the second machine learning structure, therefore, online acquisition skip execution information of the second cloud service flow when the preset online knowledge graph information of the second cloud service flow is realized is accurately determined, and online acquisition behavior configuration data used for configuring an online acquisition behavior strategy in the second cloud service flow is accurately positioned in the follow-up process.

In a separate embodiment, the feature classification structure includes a first convolution structure, a second convolution structure, a full-connected structure, and a third convolution structure; mapping the online acquisition skip execution information of the second cloud service flow to online acquisition behavior configuration data for configuring an online acquisition behavior strategy in the second cloud service flow, including: performing feature extraction on the online acquisition and migration execution information of the second cloud service process based on the first convolution structure to obtain a first convolution result corresponding to the online acquisition and migration execution information; performing feature extraction on the online acquisition skip execution information of the second cloud service flow based on a second convolution structure to obtain a second convolution result corresponding to the online acquisition skip execution information; determining a loss function of the second convolution result and the first convolution result based on the full-connected structure; and performing feature extraction on the loss function based on the third convolution structure to obtain online acquisition behavior configuration data used for configuring an online acquisition behavior strategy in the second cloud service process.

For example, after the second machine learning structure inputs the online acquisition skip execution information of the second cloud service flow into the feature classification structure, based on the learned online acquisition behavior configuration data of the online acquisition behavior policy and the map association entity between the online acquisition migration execution information and the online acquisition skip execution information of the online acquisition behavior policy, the online acquisition migration execution information of the second cloud service flow is subjected to feature extraction through the first convolution structure in the feature classification structure to obtain a first convolution result (i.e. a first convolution vector) corresponding to the online acquisition migration execution information, and the online acquisition skip execution information of the second cloud service flow is subjected to feature extraction through the second convolution structure in the feature classification structure to obtain a second convolution result (i.e. a second convolution vector) corresponding to the online acquisition skip execution information, and finally, performing feature extraction on the loss function through a third convolution structure to obtain online acquisition behavior configuration data used for configuring an online acquisition behavior strategy in a second cloud service process.

Step S103, the online acquisition behavior configuration data is applied to the online acquisition behavior strategy in the second cloud service process, so that after the key big data of the big data operation activity is acquired based on the online acquisition behavior strategy, big data mining is carried out on the key big data of the acquired big data operation activity.

For example, the big data mining service system 100 sends a configuration instruction for an online acquisition behavior policy to the big data mining service system, the big data mining service system receives the configuration instruction for the online acquisition behavior policy, trains a target machine learning network to perform prediction processing, obtains online acquisition behavior configuration data of a second cloud service process, and feeds the online acquisition behavior configuration data of the second cloud service process back to the big data mining service system 100, the big data mining service system 100 applies the online acquisition behavior configuration data to the online acquisition behavior policy in the second cloud service process through a big data registration terminal, so that online acquisition skip execution information of the first cloud service process is converted into online acquisition skip execution information of the second cloud service process, and the online acquisition skip execution information of the second cloud service process and the online acquisition migration execution information of the second cloud service process act together And then, converting the online knowledge graph information of the first cloud service process into the online knowledge graph information of the second cloud service process, thereby realizing the optimization of online acquisition objects in an automatic control online acquisition behavior strategy.

In a separate embodiment, step S104-step S105 may also be included: in step S104, a reference data set of the target machine learning network is constructed based on online knowledge map information of online acquisition objects in a plurality of reference cloud service flows in an online acquisition behavior strategy, online acquisition skip execution information of the online acquisition behavior strategy in the plurality of reference cloud service flows, and online acquisition migration execution information of the online acquisition behavior strategy in the plurality of reference cloud service flows; in step S105, the target machine learning network is trained based on the reference data set, so as to obtain the target machine learning network for online behavior configuration data collection prediction.

For example, the characteristics of the current cloud service flow (including online knowledge map information, online acquisition skip execution information, and online acquisition migration execution information) are used as the characteristics of the reference cloud service flow of the next cloud service flow. Features (including online knowledge map information, online acquisition skip execution information, and online acquisition migration execution information) in a service iteration stage of a service process of an online acquisition object in an online acquisition behavior policy may be used as a reference data set to train a target machine learning network, that is, a plurality of reference cloud service flows belong to a service iteration flow of the service process of the online acquisition object, for example, a service process cycle of the online acquisition object is 30 flows, and online acquisition behavior configuration data is optimized in units of flows, and then the number of the plurality of reference cloud service flows is 30, that is, a1 st flow, a 2 nd flow, and a … th flow 30.

For example, the plurality of reference cloud service flows may be a1 st reference cloud service flow, a 2 nd reference cloud service flow, and a 3 rd reference cloud service flow, where the 1 st reference cloud service flow is any cloud service flow in a1 st business iteration stage of the online acquisition object, the 2 nd reference cloud service flow is any cloud service flow in a 2 nd business iteration stage of the online acquisition object, and the 3 rd reference cloud service flow is any cloud service flow in a 3 rd business iteration stage of the online acquisition object.

In a separate embodiment, a reference data set of a target machine learning network is constructed based on online knowledge graph information of online acquisition objects in a plurality of reference cloud service flows in an online acquisition behavior policy, online acquisition skip execution information of the online acquisition behavior policy in the plurality of reference cloud service flows, and online acquisition migration execution information of the online acquisition behavior policy in the plurality of reference cloud service flows, and the reference data set comprises: executing the following processing aiming at any reference cloud service flow in a plurality of reference cloud service flows: acquiring online knowledge map information of a next reference cloud service flow of a reference cloud service flow, online acquisition skip execution information of the next reference cloud service flow of the reference cloud service flow and online acquisition migration execution information of the next reference cloud service flow of the reference cloud service flow; combining online knowledge map information of an online acquisition object in the online acquisition behavior strategy in a reference cloud service flow, online acquisition skip execution information of the online acquisition behavior strategy in the reference cloud service flow and online acquisition migration execution information of the reference cloud service flow into a first convolution result of the reference cloud service flow; combining online knowledge map information of a next reference cloud service flow of the reference cloud service flow, online acquisition skip execution information of the next reference cloud service flow of the reference cloud service flow and online acquisition migration execution information of the next reference cloud service flow of the reference cloud service flow into a second convolution result of the next reference cloud service flow; constructing a reference data set of a reference cloud service flow based on a first convolution result of the reference cloud service flow, on-line acquisition behavior configuration data used for configuring an on-line acquisition behavior strategy in the reference cloud service flow and a second convolution result of a next reference cloud service flow; and performing fusion processing on the reference data sets of the multiple reference cloud service processes to obtain a reference data set of the target machine learning network.

For example, when the online knowledge graph information of a certain reference cloud service flow is x1, the online collection jump execution information is y1, and the online collection migration execution information is z1, the feature vector of the reference cloud service flow is S1[ x1, y1, z1 ]. When the online knowledge graph information of the next reference cloud service flow of the reference cloud service flow is x2, the online collection jump execution information is y2, and the online collection migration execution information is z2, the feature vector of the next reference cloud service flow of the reference cloud service flow is S2[ x2, y2, z2 ]. When the online collection behavior configuration data for configuring the online collection behavior policy in the reference cloud service flow is a1, combining the online knowledge map information S1 of the reference cloud service flow, the online collection behavior configuration data for configuring the online collection behavior policy in the reference cloud service flow is a1, and the feature vector S2 of the next reference cloud service flow of the reference cloud service flow into a reference data set [ S1, a1, S2] of the reference cloud service flow. When there are N reference cloud service flows, the reference data set is [ S1, a1, S2], …, [ sN-1, aN-1, sN ], [ sN, aN, termination state ], where N is a natural number greater than 2, and the termination state represents a training termination condition.

In the process of dynamically adjusting and controlling the big data acquisition behavior of the online acquisition object in the online acquisition behavior strategy, the reference data set of the reference cloud service process can be stored in the transfer characteristic partition to collect the reference data set, so that when the number of the reference data sets of the reference cloud service process in the transfer characteristic partition reaches a set threshold value, the reference data sets of the multiple reference cloud service processes are obtained from the transfer characteristic partition, and the target machine learning network is trained based on the reference data sets of the multiple reference cloud service processes.

In a separate embodiment, training a target machine learning network based on a reference data set of a business service process of an online collection object in an online collection behavior policy to obtain the target machine learning network for online collection behavior configuration data prediction includes: constructing a weight evaluation function of the target machine learning network based on a reference data set of any reference cloud service process in the reference data set and reference evaluation information of the reference cloud service process; and optimizing parameters of the target machine learning network until the weight evaluation function is minimized, and taking the optimized parameters of the target machine learning network when the weight evaluation function is minimized as the parameters of the target machine learning network for online acquisition of behavior configuration data prediction.

In a separate embodiment, before constructing the weight evaluation function of the target machine learning network, the method further includes: training a target machine learning network to perform prediction processing based on a reference data set of any reference cloud service process in the reference data set, and obtaining prediction online acquisition behavior configuration data for configuring an online acquisition behavior strategy in the reference cloud service process; acquiring behavior configuration data on line based on prediction to obtain prediction evaluation information of a reference cloud service process; and constructing a weight evaluation function of the target machine learning network based on the reference data set of the reference cloud service flow, the prediction evaluation information of the reference cloud service flow and the reference evaluation information of the reference cloud service flow.

For example, in a training phase, a training target machine learning network performs prediction processing to obtain predicted online acquisition behavior configuration data for configuring an online acquisition behavior strategy in a reference cloud service process, and obtains predicted evaluation information of the reference cloud service process based on the predicted online acquisition behavior configuration data, for example, online knowledge map information which can be generated by an online acquisition object in the online acquisition behavior strategy is predicted based on the predicted online acquisition behavior configuration data, and online knowledge map prediction information brought by the online knowledge map information which can be generated by the online acquisition object is used as predicted evaluation information; predicting online knowledge map information which can be generated by an online acquisition object in an online acquisition behavior strategy based on the predicted online acquisition behavior configuration data, taking online knowledge map prediction information brought by the online knowledge map information which can be generated by the online acquisition object as prediction evaluation information, predicting quoted service information associated with label attributes required by the online acquisition behavior configuration data, and taking a loss function of the online knowledge map prediction information and the quoted service information as reference evaluation information. And finally, constructing a weight evaluation function of the target machine learning network based on the reference data set of the reference cloud service flow, the prediction evaluation information of the reference cloud service flow and the reference evaluation information of the reference cloud service flow.

In a separate embodiment, feature association processing is performed on online knowledge graph information of a reference cloud service process based on a first machine learning structure in a target machine learning network to obtain preset online knowledge graph information meeting big data acquisition optimization requirements in a next reference cloud service process, online acquisition behavior configuration data of an online acquisition behavior strategy and a graph association entity between online acquisition migration execution information and online acquisition skip execution information of the online acquisition behavior strategy, which are included in a feature classification structure in the target machine learning network, are mapped into predicted online acquisition behavior configuration data for configuring the online acquisition behavior strategy in the next reference cloud service process.

For example, after determining the value of the weight evaluation function (e.g., but not limited to, cross entropy loss function, etc.) of the target machine learning network based on the reference data set of any one reference cloud service flow, the predicted evaluation information of the reference cloud service flow, and the reference evaluation information of the reference cloud service flow (e.g., net value of traffic actually brought by the online collection object in the reference cloud service flow, or traffic conversion rate of the online collection object in the reference cloud service flow), it may be determined whether the value of the weight evaluation function of the target machine learning network exceeds a preset threshold, and when the value of the weight evaluation function of the target machine learning network exceeds the preset threshold, an error signal of the target machine learning network is determined based on the weight evaluation function of the target machine learning network, and the error information is propagated backwards in the target machine learning network, and optimizing the model parameters of each layer in the process of propagation.

In a separate embodiment, before constructing the weight evaluation function of the target machine learning network, the method further includes: the method comprises the steps of obtaining online knowledge graph information of a next reference cloud service process in a reference data set of the reference cloud service process, requesting an online acquisition object simulator model based on the online knowledge graph information of the next reference cloud service process to determine online knowledge graph prediction information brought by a business service process of an online acquisition object in the reference cloud service process, and taking the online knowledge graph prediction information as reference evaluation information of the reference cloud service process. For example, the actual business conversion rate of the online acquisition object in the reference cloud service flow is used as the reference evaluation information of the reference cloud service flow.

In a separate embodiment, before constructing the weight evaluation function of the target machine learning network, the method further includes: acquiring online knowledge graph information of a next reference cloud service process in a reference data set of the reference cloud service process, and requesting an online acquisition object simulator model based on the online knowledge graph information of the next reference cloud service process to determine online knowledge graph prediction information brought by a business service process of the online acquisition object in the reference cloud service process; acquiring online acquisition behavior configuration data for configuring an online acquisition behavior strategy in a reference cloud service flow; requesting an online acquisition object simulator model based on online acquisition behavior configuration data for configuring an online acquisition behavior strategy in a reference cloud service process to determine citation business information associated with a tag attribute required by online acquisition behavior configuration data, and using online knowledge map prediction information as reference evaluation information of the reference cloud service process, wherein the citation business information comprises: and taking the loss functions of the online knowledge map prediction information and the quoted service information as reference evaluation information of the reference cloud service process. For example, the net value of business actually brought by the reference cloud service flow of the object is collected on line to serve as reference evaluation information of the reference cloud service flow.

In a possible embodiment, the foregoing step S103 can be further realized by the following exemplary steps.

And step S1031, obtaining optimization information of online acquisition behavior configuration data aiming at the online acquisition behavior strategy.

In this embodiment, the big data mining service system 100 may receive optimization information sent by a front-end service node, where the optimization information is used to instruct that template content optimization is performed on currently acquired template information in the online acquisition behavior configuration data.

Step S1032, when the currently acquired template information is in the first acquisition activation mode, determining an optimization process corresponding to the currently acquired template information and all optimized service cloud computing service resources in the optimization process that are in the optimization activation state, and requesting all optimized service cloud computing service resources to respond to the optimization information, and performing template content optimization with respect to the currently acquired template information.

Step S1033, when the currently-collected template information is in the second collection activation mode, determining a currently-collected behavior configuration instruction of the currently-collected template information in the online collection behavior configuration data, and when the currently-collected behavior configuration instruction satisfies a preset optimization level, performing template content optimization with respect to the currently-collected template information.

In this embodiment, a collection template tag is stored in the current collection template information, and the collection template tag is used to indicate the state of the current collection template information; for example, when the acquisition template tag indicates that the current acquisition template information is in the first acquisition activation mode, the big data mining service system 100 determines that the current acquisition template information is in a complete big data acquisition activation state, and at this time, the big data mining service system 100 may determine an optimization process corresponding to the current acquisition template information and all optimized service cloud computing service resources in the optimization process that are in the optimization activation state.

The current acquired template information can be stored with an optimization label, and the optimization label can be used for indicating a corresponding optimization process; the big data mining service system 100 may store an optimized tag association bitmap, where the optimized tag association bitmap records a correspondence between a plurality of optimized tags and a plurality of optimized processes; the big data mining service system 100 can obtain the optimization flow corresponding to the optimization tag information stored in the current acquired template information by inquiring the optimization tag association bitmap; moreover, the big data mining service system 100 may send query information to the determined application service of the optimized process to request response information returned by all the optimized service cloud computing service resources in the optimized process.

Next, the big data mining service system 100 may determine the working state of each optimized service cloud computing service resource according to the condition that each optimized service cloud computing service resource in the optimized flow respectively returns response information; for example, the big data mining service system 100 may determine that the optimized service cloud computing service resource fed back with the response information within the set time is in the optimized activation state, and the optimized service cloud computing service resource that exceeds the set time and has no response information fed back is not in the optimized activation state; thus, after the big data mining service system 100 determines all optimized service cloud computing service resources in the optimized activated state in the optimization process, the big data mining service system 100 may request all optimized service cloud computing service resources to respond to the optimization request, and perform template content optimization with respect to the currently acquired template information.

In addition, when the acquisition template tag indicates that the currently acquired template information is in the second acquisition activation mode, the big data mining service system 100 determines that the currently acquired template information is in a partial big data acquisition activation state, that is, the currently acquired template information contains some data which is not data acquisition activated, at this time, the big data mining service system 100 can determine a currently acquired behavior configuration instruction of the currently acquired template information in the online acquisition behavior configuration data, and look up a pre-configured optimization level policy table, wherein the optimization level policy table records the optimization levels corresponding to the multiple signaling; when the big data mining service system 100 determines the optimization level corresponding to the current collection behavior configuration instruction by searching the optimization level policy table, and when the big data mining service system 100 determines that the optimization level corresponding to the current collection behavior configuration instruction meets the preset optimization level, the big data mining service system 100 determines that the current collection template information meets the optimization requirement, and then the big data mining service system 100 performs template content optimization on the current collection template information.

It can be seen that in the solution provided by the present disclosure, optimization information of online collection behavior configuration data for an online collection behavior policy is obtained, where the optimization information is used to instruct that template content optimization is performed on current collection template information in the online collection behavior configuration data; under the condition that the current acquisition template information is in a first acquisition activation mode, determining an optimization process corresponding to the current acquisition template information and all optimized service cloud computing service resources in an optimized activation state in the optimization process, requesting all optimized service cloud computing service resources to respond to optimization information, and executing template content optimization aiming at the current acquisition template information; in addition, under the condition that the current acquisition template information is in the second acquisition activation mode, a current acquisition behavior configuration instruction of the current acquisition template information in the online acquisition behavior configuration data is determined, and when the current acquisition behavior configuration instruction meets a preset optimization level, the template content optimization is executed aiming at the current acquisition template information. Therefore, the current acquisition template information is judged to be in the first acquisition activation mode or the second acquisition activation mode, the optimization information of the current acquisition template information is responded to different conditions, the template content optimization is executed, the reliability of big data acquisition of the acquisition template information is improved, and data pollution is avoided.

In an independent embodiment, for different working states of the optimization process, after the step S1031 is executed, the big data mining service system 100 may further obtain the big data collection description information of the online collection behavior configuration data.

Next, the big data mining service system 100 may obtain the working state of the optimization process in the above manner; moreover, when the instruction indicating that the optimization process is in the optimization operation process is obtained, the big data mining service system 100 may switch the current collection template information from the first collection activation mode to the second collection activation mode, obtain the current collection behavior configuration instruction of the current collection template information in the online collection behavior configuration data according to the big data collection description information, and then execute step S1033.

In this embodiment, the current collection behavior configuration instruction may be recorded in a specified field of the big data collection description information.

In addition, the big data mining service system 100 may switch the current acquisition template information from the second acquisition activation mode to the first acquisition activation mode when acquiring the information indicating that the optimization process is in the optimization non-maintenance mode, and acquire the description distribution in the online acquisition behavior configuration data according to the big data acquisition description information; thereby performing step S1032.

In a separate embodiment, when step S1033 is executed to determine the current collecting behavior configuration instruction of the current collecting template information in the online collecting behavior configuration data, the following scheme may be adopted by the big data mining service system 100:

first, the big data mining service system 100 may use a big data acquisition configuration interface of the current acquisition template information in the online acquisition behavior configuration data as a target configuration interface, and determine a traversal configuration interface service interval using a preset big data acquisition configuration interface service interval in the current acquisition template information as a fixed range. In this embodiment, the preset big data acquisition configuration interface service interval may be recorded in the preset configuration interface service interval in the current acquisition template information. Next, the big data mining service system 100 may determine a configured interface high frequency interval of the currently acquired template information from the traversal configured interface service interval according to the configured interface frequency interval of the currently acquired template information; the high-frequency range of the configuration interface is a high-temperature range for storing the configuration command of the acquisition behavior in the current acquisition template information. In this embodiment, the configuration interface frequency interval may be stored in the metadata of the current collection template information. The big data mining service system 100 may then search the configuration interface high frequency interval for the current collection behavior configuration instructions of the current collection template information. Next, when the current acquisition behavior configuration instruction is not searched in the configuration interface high-frequency interval, the big data mining service system 100 may use the big data acquisition configuration interface of the current acquisition template information in the online acquisition behavior configuration data as a target configuration interface, use the acquisition template tag attribute corresponding to the current acquisition template information as a target tag attribute, and divide according to the set tag attribute to obtain the tag attributes of the plurality of acquisition template information. Then, the big data mining service system 100 may obtain a tag attribute sequence corresponding to the tag attribute of each of the collected template information in the tag attributes of the collected template information, and fuse all the tag attribute sequences according to the order of the tag attributes of the collected template information corresponding to each of the tag attribute sequences, so as to generate a current collection behavior configuration instruction of the current collected template information.

In addition, in a separate embodiment, in order to evaluate the overall importance of the online collection behavior policy, for the current collection template information corresponding to the online collection behavior policy, the information processing method may further include the following steps:

first, the big data mining service system 100 may obtain historical big data collection behavior information of the current collection template information in a preset big data collection cycle and big data collection behavior feedback information corresponding to each historical big data collection behavior information. In this embodiment, the preset big data acquisition cycle period may be five cycle periods of the online acquisition behavior strategy in the past; the historical big data collection behavior information may include big data control information during the big data collection behavior within each preset big data collection cycle period. Next, the big data mining service system 100 may perform big data acquisition index change calculation on the currently acquired template information according to the historical big data acquisition behavior information and the big data acquisition behavior feedback information to obtain big data acquisition index change evaluation information corresponding to the currently acquired template information in each big data acquisition cycle period. In this embodiment, the big data mining service system 100 may perform service index increase calculation on the currently acquired template information according to the historical big data acquisition behavior information and the big data acquisition behavior feedback information by using a least square fitting, for example, so as to obtain the big data acquisition index change evaluation information corresponding to the currently acquired template information in each big data acquisition cycle. Then, the big data mining service system 100 may perform importance parameter calculation on each big data acquisition index change evaluation information, respectively, to obtain at least one group of big data acquisition index importance parameters. Next, the big data mining service system 100 may determine current big data acquisition index change evaluation information to be estimated and related big data acquisition index change evaluation information of a period adjacent to the current big data acquisition index change evaluation information; the relevant big data acquisition index change evaluation information is big data acquisition index change evaluation information of a previous period adjacent to the current big data acquisition index change evaluation information. Then, the big data mining service system 100 may determine, according to the big data acquisition index importance parameter, an expected big data acquisition effective transformation parameter corresponding to each of the current big data acquisition index change evaluation information and the relevant big data acquisition index change evaluation information. Next, the big data mining service system 100 may perform big data collection importance prediction on the currently collected template information according to all the expected big data collection effective transformation parameters and the big data collection index change evaluation information corresponding to each of the expected big data collection effective transformation parameters, so as to obtain the expected big data collection importance data corresponding to the currently collected template information. Then, the big data mining service system 100 may obtain historical big data collection importance data corresponding to the historical collection template information, and determine current big data collection expected data of the current collection template information according to the historical big data collection importance data; and the historical acquisition template information is the prior acquisition template information corresponding to the current acquisition template information. Next, the big data mining service system 100 may perform big data collection importance parameter calculation on the current big data collection expected data according to the expected big data collection importance data to obtain the big data collection importance influence parameter of the current collection template information. Then, the big data mining service system 100 may perform importance prediction on the currently acquired template information according to the big data acquisition importance influencing parameter, so as to obtain the reference big data acquisition importance corresponding to the currently acquired template information.

Therefore, based on the scheme provided by the embodiment of the disclosure, the importance of the current acquisition template information is predicted, and the overall importance of the current acquisition template information of the online acquisition behavior strategy can be evaluated based on the estimated importance of the large data acquisition, so that the indexes of the online acquisition behavior strategy are objectively evaluated, and the reliability of service expansion is improved.

As a possible implementation manner, in order to improve reliability when acquiring the historical big data acquisition importance data and prevent data from being leaked, the big data mining service system 100 may adopt the following manner when acquiring the historical big data acquisition importance data corresponding to the historical acquisition template information:

firstly, the big data mining service system 100 may determine a target historical acquisition content optimization item corresponding to the current acquisition template information, and obtain a historical acquisition content optimization data packet stored by the target historical acquisition content optimization item; the historical acquisition content optimization data packet is a data packet of all historical acquisition template information stored in a target historical acquisition content optimization project. Next, the big data mining service system 100 may obtain the optimized set tag attributes of the history collected content optimized data packet fed back by the target history collected content optimized item. The big data mining service system 100 may then decode the historical acquisition content optimization data packet using the optimization set tag attributes. Next, the big data mining service system 100 may acquire an acquisition flow start node and an acquisition flow end node of the historical acquisition template information. Then, the big data mining service system 100 may determine a target collection process according to the collection process start node and the collection process end node. Next, the big data mining service system 100 may obtain target historical big data collection subscription information of the decoded historical collection content optimization data packet in the target collection process. Then, the big data mining service system 100 may analyze historical big data collection importance data corresponding to the current collection template information from the target historical big data collection subscription information.

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

The obtaining module 310 is configured to obtain online knowledge map information of an online collecting object of an online collecting tool of the big data registration terminal in the online collecting behavior policy in the first cloud service flow, online collecting skip execution information of the online collecting behavior policy in the first cloud service flow, and online collecting migration execution information of the online collecting behavior policy in the first cloud service flow.

The training module 320 is configured to train the target machine learning network based on the online knowledge map information of the first cloud service flow, the online collection skip execution information of the first cloud service flow, and the online collection migration execution information of the first cloud service flow, so as to obtain online collection behavior configuration data for configuring an online collection behavior policy in a second cloud service flow, where the second cloud service flow is the migration cloud service flow of the first cloud service flow.

The application module 330 is configured to apply the online acquisition behavior configuration data to the online acquisition behavior policy in the second cloud service process, so as to perform big data mining on the acquired key big data of the big data operation activity after performing the key big data acquisition of the big data operation activity based on the online acquisition behavior policy.

Fig. 4 illustrates a hardware structure diagram of a big data mining service system 100 for implementing the above-mentioned big data mining method based on cloud computing services according to an embodiment of the present disclosure, and as shown in fig. 4, the big data mining service system 100 may include a processor 110, a machine-readable storage medium 120, a bus 130, and a transceiver 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 computing service-based big data mining method according to the above method embodiment, the processor 110, the machine-readable storage medium 120, and the transceiver 140 are connected through the bus 130, and the processor 110 may be configured to control transceiving actions of the transceiver 140, so as to perform data transceiving with the big data registration 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 mining service system 100, which implement principles and technical effects are similar, and details of this embodiment are not described herein again.

In addition, the embodiment of the disclosure also provides a readable storage medium, wherein the readable storage medium is preset with a computer execution instruction, and when a processor executes the computer execution instruction, the cloud computing service-based big data mining method 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. Thus, in a separate embodiment, not limiting, alternative configurations of the embodiments of the present specification are contemplated as being consistent with the teachings of the present specification. Accordingly, the embodiments of the present description are not limited to only those embodiments explicitly described and depicted herein.

Claims

1. A big data mining method based on cloud computing service is applied to a big data mining service system, the big data mining service system is in communication connection with a plurality of big data registration terminals, and the method comprises the following steps:

2. The cloud computing service-based big data mining method according to claim 1, wherein the step of training a target machine learning network based on the online knowledge graph information of the first cloud service process, the online collection skip execution information of the first cloud service process, and the online collection migration execution information of the first cloud service process to obtain online collection behavior configuration data for configuring the online collection behavior policy in a second cloud service process includes:

performing the following processing based on the target machine learning network:

determining preset online knowledge graph information meeting big data acquisition optimization requirements in the second cloud service flow based on the online knowledge graph information of the first cloud service flow, and determining online acquisition skip execution information of the second cloud service flow when the preset online knowledge graph information is realized;

and determining online acquisition behavior configuration data used for configuring the online acquisition behavior strategy in the second cloud service flow based on a linkage relation that the online acquisition migration execution information of the second cloud service flow and online acquisition behavior configuration data used for configuring the online acquisition behavior strategy in the second cloud service flow jointly act on the online acquisition skip execution information of the second cloud service flow.

3. The cloud computing service-based big data mining method according to claim 2, wherein the target machine learning network comprises a first machine learning structure and a feature classification structure;

the step of determining preset online knowledge graph information meeting big data acquisition optimization requirements in the second cloud service process based on the online knowledge graph information of the first cloud service process comprises the following steps:

performing feature association processing on the online knowledge graph information of the first cloud service process based on the first machine learning structure to obtain preset online knowledge graph information meeting big data acquisition optimization requirements in the second cloud service process;

the step of determining online collection behavior configuration data for configuring the online collection behavior policy in the second cloud service process includes:

and mapping the online acquisition jump execution information of the second cloud service flow into online acquisition behavior configuration data used for configuring the online acquisition behavior strategy in the second cloud service flow based on the characteristic association relation between the online acquisition behavior configuration data and the online acquisition migration execution information of the online acquisition behavior strategy, which are included in the characteristic classification structure, and the online acquisition jump execution information of the online acquisition behavior strategy.

4. The cloud computing service-based big data mining method according to claim 3, wherein the target machine learning network further comprises a second machine learning structure;

the step of determining the online acquisition skip execution information of the second cloud service process when the preset online knowledge graph information is realized includes:

based on the feature association relationship between the online acquisition jump execution information of the online acquisition behavior strategy and the online knowledge graph information of the online acquisition object, which is included in the second machine learning structure, performing feature association processing on the preset online knowledge graph information of the second cloud service process to obtain online acquisition jump execution information of the second cloud service process when the preset online knowledge graph information of the second cloud service process is realized;

or based on the feature association relation between the online acquisition jump execution information of the online acquisition behavior strategy and the online knowledge graph information change of the online acquisition object, which is included by the second machine learning structure, the online knowledge graph information of the first cloud service flow and the loss function of the preset online knowledge graph information of the second cloud service flow are subjected to feature conversion processing, and the online acquisition jump execution information of the second cloud service flow is obtained when the preset online knowledge graph information of the second cloud service flow is realized.

5. The cloud computing service-based big data mining method according to claim 3, wherein the feature classification structure comprises a first convolution structure, a second convolution structure, a full-connected structure and a third convolution structure;

the step of mapping the online acquisition skip execution information of the second cloud service flow to online acquisition behavior configuration data for configuring the online acquisition behavior policy in the second cloud service flow includes:

performing feature extraction on the online acquisition and migration execution information of the second cloud service process based on the first convolution structure to obtain a first convolution result corresponding to the online acquisition and migration execution information;

performing feature extraction on the online acquisition skip execution information of the second cloud service process based on the second convolution structure to obtain a second convolution result corresponding to the online acquisition skip execution information;

determining a loss function of the second convolution result and the first convolution result based on the full-connected structure;

and performing feature extraction on the loss function based on the third convolution structure to obtain online acquisition behavior configuration data for configuring the online acquisition behavior strategy in the second cloud service process.

6. The cloud computing service-based big data mining method according to claim 1, wherein the method further comprises:

constructing a reference data set of the target machine learning network based on online knowledge map information of online acquisition objects in a plurality of reference cloud service flows in the online acquisition behavior strategy, online acquisition skip execution information of the online acquisition behavior strategy in the plurality of reference cloud service flows, and online acquisition migration execution information of the online acquisition behavior strategy in the plurality of reference cloud service flows;

and training the target machine learning network based on the reference data set to obtain the target machine learning network for predicting online collected behavior configuration data.

7. The cloud computing service-based big data mining method according to claim 6, wherein the step of constructing the reference data set of the target machine learning network based on online knowledge graph information of online collected objects in the online collection behavior policy in a plurality of reference cloud service flows, online collection skip execution information of the online collection behavior policy in the plurality of reference cloud service flows, and online collection migration execution information of the online collection behavior policy in the plurality of reference cloud service flows comprises:

executing the following processing for any reference cloud service flow in the plurality of reference cloud service flows:

acquiring online knowledge map information of a next reference cloud service flow of the reference cloud service flow, online acquisition skip execution information of the next reference cloud service flow of the reference cloud service flow and online acquisition migration execution information of the next reference cloud service flow of the reference cloud service flow;

combining online knowledge map information of online acquisition objects in the online acquisition behavior strategy in the reference cloud service flow, online acquisition skip execution information of the online acquisition behavior strategy in the reference cloud service flow, and online acquisition migration execution information of the reference cloud service flow into a first convolution result of the reference cloud service flow;

combining online knowledge map information of a next reference cloud service flow of the reference cloud service flow, online acquisition skip execution information of the next reference cloud service flow of the reference cloud service flow, and online acquisition migration execution information of the next reference cloud service flow of the reference cloud service flow into a second convolution result of the next reference cloud service flow;

constructing a reference data set of the reference cloud service flow based on a first convolution result of the reference cloud service flow, on-line acquisition behavior configuration data used for configuring the on-line acquisition behavior policy in the reference cloud service flow, and a second convolution result of the next reference cloud service flow;

and fusing the reference data sets of the reference cloud service processes to obtain the reference data set of the target machine learning network.

8. The cloud computing service-based big data mining method according to any one of claims 1 to 7, wherein the step of applying the online collection behavior configuration data to the online collection behavior policy in the second cloud service process includes:

acquiring optimization information of online acquisition behavior configuration data aiming at the online acquisition behavior strategy in the second cloud service process; wherein the optimization information is used to indicate that template content optimization is performed in the online collection behavior policy for current collection template information in online collection behavior configuration data;

when the current acquisition template information is in a first acquisition activation mode, determining an optimization process corresponding to the current acquisition template information and all optimized service cloud computing service resources in an optimized activation state in the optimization process, requesting all optimized service cloud computing service resources to respond to the optimization information, and executing template content optimization aiming at the current acquisition template information;

when the current acquisition template information is in a second acquisition activation mode, determining a current acquisition behavior configuration instruction of the current acquisition template information in online acquisition behavior configuration data, and when the current acquisition behavior configuration instruction meets a preset optimization level, executing template content optimization aiming at the current acquisition template information.

9. The cloud computing service-based big data mining method according to claim 1, wherein after the step of obtaining optimization information of online collection behavior configuration data for the online collection behavior policy, the method further comprises:

acquiring big data acquisition description information of the online acquisition behavior configuration data;

when the optimization flow is indicated to be in the optimization operation flow, the current acquisition template information is switched from a first acquisition activation mode to a second acquisition activation mode, and a current acquisition behavior configuration instruction of the current acquisition template information in the online acquisition behavior configuration data is acquired according to big data acquisition description information;

and when the current acquisition template information is acquired and used for indicating that the optimization process is in an updating non-maintenance mode, switching the current acquisition template information from a second acquisition activation mode to a first acquisition activation mode, and acquiring the description distribution in the online acquisition behavior configuration data according to the big data acquisition description information.

10. A big data mining service system, characterized in that the big data mining service system comprises:

a memory for storing executable instructions;

a processor, configured to execute the executable instructions stored in the memory, and implement the cloud computing service-based big data mining method according to any one of claims 1 to 9.