CN113535798A

CN113535798A - Network training method based on big data mining and digital content center

Info

Publication number: CN113535798A
Application number: CN202110611130.9A
Authority: CN
Inventors: 夏红梅
Original assignee: Individual
Current assignee: Individual
Priority date: 2020-12-17
Filing date: 2020-12-17
Publication date: 2021-10-22
Also published as: CN112506999A; CN112506999B; CN113535799A

Abstract

The embodiment of the application provides a network training method based on big data mining and a digital content center, wherein at least one label mining source, original label mining source data corresponding to the at least one label mining source respectively, updated reference label data corresponding to the at least one label mining source respectively and a standard mining result corresponding to the at least one label mining source respectively are obtained, and then constraint training is carried out based on the standard mining results until a constraint training process meets a target training end condition, so that a target correction network and a target mining network are obtained.

Description

Network training method based on big data mining and digital content center

The application is a divisional application of Chinese application with the name of 'cloud computing and artificial intelligence based big data mining method and digital content center', which is invented and created by application number 202011499130.6 and has the application date of 12-17.2020.

Technical Field

The application relates to the technical field of big data mining optimization, in particular to a network training method based on big data mining and a digital content center.

Background

The big data mining strategy refers to a strategy for searching information hidden in a large amount of big data through an algorithm. The method has the advantages that the big data mining strategy in the mining source (such as a certain application program or a certain terminal device) is updated, and key data can be provided for implementation in the subsequent deep learning-based mining process.

In the related art, a big data mining strategy in a mining source is generally updated manually by a big data developer, so that the updating efficiency is low, and the reliability of an updating result of the obtained big data mining strategy is poor.

Disclosure of Invention

In order to overcome at least the above disadvantages in the prior art, the present application aims to provide a network training method and a digital content center based on big data mining, which automatically obtain a target mining result corresponding to a target mining source based on original mining source data and updated user behavior data corresponding to the target mining source, and then automatically obtain a global update result of a target big data mining strategy in the target mining source based on the target mining result. Based on the process, the big data mining strategy can be automatically updated, the updating process of the big data mining strategy does not depend on manpower, the efficiency of updating the big data mining strategy in the mining source is improved, and the reliability of the updating result of the obtained big data mining strategy is high.

In a first aspect, the present application provides a cloud computing and artificial intelligence based big data mining method, which is applied to a digital content center, where the digital content center is in communication connection with a plurality of digital content subscription devices, and the method includes:

acquiring an acquisition configuration result which is output by acquiring and configuring acquisition feature items, wherein the acquisition configuration result comprises a target big data mining strategy aiming at each acquisition feature item;

acquiring a target mining source corresponding to the target big data mining strategy, original mining source data corresponding to the target mining source and updated user behavior data corresponding to the target mining source, wherein the updated user behavior data is determined based on at least partial update results of the target big data mining strategy in the target mining source;

calling a target mining network, and acquiring a target mining result corresponding to the target mining source based on the original mining source data and the updated user behavior data, wherein the target mining result is used for indicating target attributes of all mining objects in the target mining source, and the target attribute of any mining object is used for indicating that any mining object belongs to the target big data mining strategy or that any mining object does not belong to the target big data mining strategy;

and updating the target big data mining strategy in the target mining source based on the target mining result to obtain a global updating result of the target big data mining strategy in the target mining source, and performing big data mining based on the global updating result.

In a possible implementation manner of the first aspect, the invoking a target mining network, and obtaining a target mining result corresponding to the target mining source based on the original mining source data and the updated user behavior data includes:

calling a target mining network, and sequentially executing forward mining analysis of a first iteration strategy based on the original mining source data and the associated feature data of the updated user behavior data to obtain a first mining component corresponding to the target mining source;

sequentially executing negative mining analysis of the first iteration strategy based on the target distinguishing mining characteristics corresponding to the first mining component to obtain a second mining component corresponding to the target mining source;

and carrying out target distinguishing mining analysis on the second mining component to obtain a target mining result corresponding to the target mining source.

In a possible implementation manner of the first aspect, the first iteration strategy includes a preset number of iterations mining, the preset number is three, and any one positive mining analysis includes one differential mining analysis and one frequent pattern mining analysis;

the sequentially executing forward mining analysis of a first iteration strategy based on the associated feature data of the original mining source data and the updated user behavior data to obtain a first mining component corresponding to the target mining source includes:

performing first distinguishing mining analysis on the associated feature data of the original mining source data and the updated user behavior data to obtain a first distinguishing mining feature corresponding to the target mining source;

performing first frequent pattern mining analysis on the first distinguishing mining features to obtain first frequent pattern mining features corresponding to the target mining source;

performing second distinguishing mining analysis on the first frequent pattern mining characteristics to obtain second distinguishing mining characteristics corresponding to the target mining source;

performing second frequent pattern mining analysis on the second distinguishing mining features to obtain second frequent pattern mining features corresponding to the target mining source;

performing third region mining analysis on the second frequent pattern mining characteristics to obtain third region mining characteristics corresponding to the target mining source;

and carrying out third frequent pattern mining analysis on the third region mining characteristics to obtain a first mining component corresponding to the target mining source.

In one possible implementation manner of the first aspect, any one negative mining analysis includes a primary inverse differential mining analysis and a primary differential mining analysis; the sequentially executing negative mining analysis of the first iteration strategy based on the target distinguishing mining characteristics corresponding to the first mining component to obtain a second mining component corresponding to the target mining source comprises:

performing first inverse differential mining analysis on the target differential mining characteristics corresponding to the first mining component to obtain first negative mining characteristics corresponding to the target mining source;

performing fourth differential excavation analysis on the splicing feature of the first negative excavation feature and the third differential excavation feature to obtain a fourth differential excavation feature corresponding to the target excavation source;

performing second inverse differential excavation analysis on the fourth differential excavation feature to obtain a second negative excavation feature corresponding to the target excavation source;

performing fifth distinguishing mining analysis on the second negative mining feature and the splicing feature of the second distinguishing mining feature to obtain a fifth distinguishing mining feature corresponding to the target mining source;

performing third inverse differential excavation analysis on the fifth differential excavation feature to obtain a third negative excavation feature corresponding to the target excavation source;

and carrying out sixth distinguishing mining analysis on the third negative mining characteristic and the splicing characteristic of the first distinguishing mining characteristic to obtain a second mining component corresponding to the target mining source.

In a possible implementation manner of the first aspect, after the invoking of the target mining network and obtaining the target mining result corresponding to the target mining source based on the original mining source data and the updated user behavior data, the method further includes:

calling a target correction network, and acquiring target correction reference degree information based on the original mining source data and the target mining result;

the target correction network comprises at least one distinguishing mining network unit, at least one fusion network unit and a correction classification network unit which are sequentially connected;

the calling of the target correction network and the obtaining of the target correction reference degree information based on the original mining source data and the target mining result comprise:

inputting the original mining source data and the target mining result into a first differentiated mining network unit in the target correction network for processing to obtain differentiated mining characteristics output by the first differentiated mining network unit;

from the second differentiated mining network unit, inputting the differentiated mining features output by the previous differentiated mining network unit into the next differentiated mining network unit for processing to obtain the differentiated mining features output by the next differentiated mining network unit;

inputting the distinguishing mining characteristics output by the last distinguishing mining network unit into a first converged network unit for processing to obtain the converged characteristics output by the first converged network unit;

from the second converged network unit, inputting the converged features output by the previous converged network unit into the next converged network unit for processing to obtain the converged features output by the next converged network unit;

and inputting the fusion characteristics output by the last fusion network unit into the correction classification network unit for processing to obtain the target correction reference degree information output by the correction classification network unit.

For example, in one possible implementation of the first aspect, the target mining network is obtained by:

acquiring at least one marker mining source, original marker mining source data corresponding to the at least one marker mining source respectively, updated reference marker data corresponding to the at least one marker mining source respectively and standard mining results corresponding to the at least one marker mining source respectively;

and performing supervision training on the initial mining network based on the original tag mining source data corresponding to the at least one tag mining source, the updated reference tag data corresponding to the at least one tag mining source and the standard mining result corresponding to the at least one tag mining source to obtain the target mining network.

In a possible implementation manner of the first aspect, before the invoking the target mining network and obtaining the target mining result corresponding to the target mining source based on the original mining source data and the updated user behavior data, the method further includes:

calling the initial mining network, and acquiring a prediction mining result corresponding to a first label mining source in the at least one label mining source based on original label mining source data corresponding to the first label mining source and updated reference label data corresponding to the first label mining source;

calling the initial correction network, and acquiring first correction reference degree information based on original tag mining source data corresponding to the first tag mining source and a prediction mining result corresponding to the first tag mining source;

acquiring second correction reference degree information based on original tag mining source data corresponding to the first tag mining source and a standard mining result corresponding to the first tag mining source;

determining first difference information based on the first correction reference degree information and the second correction reference degree information;

updating parameters of the initial correction network based on the first difference information;

responding to the updating process of the parameters of the initial correction network to meet a first training end condition, and obtaining a first correction network;

calling the initial mining network, and acquiring a prediction mining result corresponding to a second label mining source based on original label mining source data corresponding to the second label mining source in at least one first label mining source and updated reference label data corresponding to the second label mining source;

calling the first correction network, and acquiring third correction reference degree information based on original tag mining source data corresponding to the second tag mining source and a prediction mining result corresponding to the second tag mining source;

determining second difference information based on the third correction reference degree information, the predicted mining result corresponding to the second label mining source and the standard mining result corresponding to the second label mining source;

updating parameters of the initial mining network based on the second difference information;

responding to the updating process of the parameters of the initial mining network to meet a second training end condition, and obtaining a first mining network;

and in response to that the constraint training process does not meet the target training end condition, continuing to perform constraint training on the first correction network and the first mining network until the constraint training process meets the target training end condition, and obtaining the target correction network and the target mining network.

In a possible implementation manner of the first aspect, the target mining source is a starting at least partial mining source corresponding to the target big data mining policy in an initial mining source; after the target big data mining strategy is updated in the target mining source based on the target mining result to obtain a global updating result of the target big data mining strategy in the target mining source, the method further includes:

in response to that the global update result of the target big data mining strategy in the target mining source does not meet the update training end condition, acquiring a next at least partial mining source corresponding to the target big data mining strategy in the initial mining source based on the global update result of the target big data mining strategy in the target mining source;

obtaining a global updating result of the target big data mining strategy in the next at least partial mining source;

and in response to the global update result of the target big data mining strategy in the next at least partial mining source meeting the update training end condition, acquiring the global update result of the target big data mining strategy in the initial mining source based on the acquired global update result of the target big data mining strategy in each at least partial mining source.

In a possible implementation manner of the first aspect, the target mining source is obtained from a service subscription mining source that includes a target mining node corresponding to the target big data mining policy;

the target attribute of any mining object is used for indicating that any mining object belongs to the target mining node or that any mining object does not belong to the target mining node;

the updating the target big data mining strategy in the target mining source based on the target mining result to obtain a global updating result of the target big data mining strategy in the target mining source includes:

determining a target mining object belonging to the target mining node in each mining object in the target mining source based on the target mining result;

based on the target excavation object, marking excavation service segments of the target excavation nodes and an incidence relation between the excavation service segments of the target excavation nodes in the target excavation source to obtain a target marking result;

and acquiring a global updating result of the target mining node in the target mining source based on the target marking result.

In a possible implementation manner of the first aspect, the step of obtaining an acquisition configuration result of the acquisition configuration output of the acquisition feature item includes:

acquiring a service function area acquisition process obtained by a current information push source based on an information push strategy;

acquiring large data label information acquired by a plurality of key operation acquisition nodes in key acquisition node operation distribution of the service functional area acquisition process; each key operation acquisition node is used for representing one or more acquisition page objects which need to be acquired in the acquisition preparation process of the service functional area, and the acquisition feature items of the acquisition page objects represented by each key operation acquisition node need to be acquired and activated; the collected big data label information of any key operation collecting node is used for reflecting the collection type relation between the any key operation collecting node and other key operation collecting nodes;

according to the collected big data label information of each key operation collection node, layering at least two key operation collection node services into a target service layered collection plan, wherein the target service layered collection plan is used for indicating collection configuration of collection characteristic items of collection page objects represented by the key operation collection nodes of the service layers;

updating the running distribution of the key acquisition nodes by adopting the target business layered acquisition plan, sending the updated running distribution of the key acquisition nodes to an acquisition configuration process of a software acquisition plan, indicating the acquisition configuration process of the software acquisition plan according to the indication of the target business layered acquisition plan, acquiring and configuring the acquisition characteristic items of the acquisition page objects represented by the business layered key operation acquisition nodes in the acquisition preparation process of the business functional area acquisition process, and outputting the acquisition configuration result.

In a second aspect, an embodiment of the present application further provides a cloud computing and artificial intelligence-based big data mining apparatus, which is applied to a digital content center, where the digital content center is communicatively connected to a plurality of digital content subscription devices, and the apparatus includes:

the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring an acquisition configuration result which is output by acquiring and configuring acquisition characteristic items, and the acquisition configuration result comprises a target big data mining strategy aiming at each acquisition characteristic item;

a second obtaining module, configured to obtain a target mining source corresponding to the target big data mining policy, original mining source data corresponding to the target mining source, and updated user behavior data corresponding to the target mining source, where the updated user behavior data is determined and obtained based on at least a partial update result of the target big data mining policy in the target mining source;

the calling module is used for calling a target mining network, and acquiring a target mining result corresponding to the target mining source based on the original mining source data and the updated user behavior data, wherein the target mining result is used for indicating target attributes of all mining objects in the target mining source, and the target attribute of any mining object is used for indicating that any mining object belongs to the target big data mining strategy or that any mining object does not belong to the target big data mining strategy;

and the updating module is used for updating the target big data mining strategy in the target mining source based on the target mining result to obtain a global updating result of the target big data mining strategy in the target mining source, and mining big data based on the global updating result.

In a third aspect, an embodiment of the present application further provides a cloud computing and artificial intelligence based big data mining system, where the cloud computing and artificial intelligence based big data mining system includes a digital content center and a plurality of digital content subscription devices communicatively connected to the digital content center;

the digital content center is configured to:

In a fourth aspect, an embodiment of the present application further provides a digital content center, where the digital content center includes a processor, a machine-readable storage medium, and a network interface, where the machine-readable storage medium, the network interface, and the processor are connected through a bus system, the network interface is configured to be communicatively connected to at least one digital content subscription device, the machine-readable storage medium is configured to store a program, an instruction, or code, and the processor is configured to execute the program, the instruction, or the code in the machine-readable storage medium to perform a cloud computing and artificial intelligence based big data mining method in the first aspect or any possible implementation manner of the first aspect.

In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, where instructions are stored in the computer-readable storage medium, and when the instructions are executed, the computer executes the cloud computing and artificial intelligence based big data mining method in the first aspect or any one of the possible implementation manners of the first aspect.

Based on any one of the aspects, the target mining result corresponding to the target mining source is automatically obtained based on the original mining source data and the updated user behavior data corresponding to the target mining source, and then the global updating result of the target big data mining strategy in the target mining source is automatically obtained based on the target mining result. Based on the process, the big data mining strategy can be automatically updated, the updating process of the big data mining strategy does not depend on manpower, the efficiency of updating the big data mining strategy in the mining source is improved, and the reliability of the updating result of the obtained big data mining strategy is high.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that need to be called in the embodiments are briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.

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

fig. 2 is a schematic flowchart of a cloud computing and artificial intelligence based big data mining method according to an embodiment of the present application;

fig. 3 is a functional module schematic diagram of a cloud computing and artificial intelligence based big data mining device according to an embodiment of the present application;

fig. 4 is a schematic block diagram of structural components of a digital content center for implementing the cloud computing and artificial intelligence based big data mining method according to an embodiment of the present application.

Detailed Description

The present application will now be described in detail with reference to the drawings, and the specific operations in the method embodiments may also be applied to the apparatus embodiments or the system embodiments.

Fig. 1 is an interaction diagram of a cloud computing and artificial intelligence based big data mining system 10 according to an embodiment of the present application. The cloud computing and artificial intelligence based big data mining system 10 may include a digital content center 100 and a digital content subscription device 200 communicatively connected with the digital content center 100. The cloud computing and artificial intelligence based big data mining system 10 shown in fig. 1 is only one possible example, and in other possible embodiments, the cloud computing and artificial intelligence based big data mining system 10 may also include only a part of the components shown in fig. 1 or may also include other components.

Based on the inventive concept of the technical solution provided by the present application, the digital content center 100 provided by the present application may be applied to a scenario where a big data technology or a cloud computing technology may be applied, such as smart medical, smart city management, smart industrial internet, and general service monitoring management, and may further be applied to, but not limited to, new energy vehicle system management, smart cloud office, cloud platform data processing, cloud game data processing, cloud live broadcast processing, cloud vehicle management platform, block chain financial data service platform, and the like.

In this embodiment, the digital content center 100 and the digital content subscription device 200 in the cloud computing and artificial intelligence based big data mining system 10 may cooperatively perform the cloud computing and artificial intelligence based big data mining method described in the following method embodiment, and the detailed description of the following method embodiment may be referred to in the steps of the digital content center 100 and the digital content subscription device 200.

In order to solve the technical problem in the foregoing background art, fig. 2 is a schematic flowchart of a cloud computing and artificial intelligence based big data mining method provided in an embodiment of the present application, and the cloud computing and artificial intelligence based big data mining method provided in this embodiment may be executed by the digital content center 100 shown in fig. 1, and the cloud computing and artificial intelligence based big data mining method is described in detail below.

And step S110, acquiring an acquisition configuration result for acquiring, configuring and outputting the acquisition characteristic items.

In this embodiment, the acquisition configuration result may specifically include a target big data mining policy for each acquisition feature item. In detail, in some alternative examples, the collection feature item may be configured according to the requirement of the actual software function, and may specifically be customized, or refer to a conventional collection configuration task in the prior art, which is not limited herein. For example, the collection characteristic item may be each business function item under a certain big data collection template, such as an order behavior collection characteristic item, a login behavior collection characteristic item, and the like, which is not specifically limited herein. In addition, the target big data mining strategy may refer to a way, a scheme, or logic of mining big data for the collected feature item, for example, for the collected feature item of the order behavior, a plurality of mining objects may be arranged based on the feature of the order behavior, so that the big data of the order behavior may be mined subsequently based on the plurality of mining objects of the order behavior.

Step S120, a target mining source corresponding to the target big data mining strategy, original mining source data corresponding to the target mining source and updated user behavior data corresponding to the target mining source are obtained.

In this embodiment, the updated user behavior data may be determined based on at least a partial update result of the target big data mining policy in the target mining source, and may be used to characterize an update change of the user behavior.

In this embodiment, the target mining source may refer to a business event source associated with the target big data mining policy, for example, for the target big data mining policy corresponding to the order behavior collection feature item, the target mining source may refer to an order placing data source, an order collection data source, and the like, but is not limited thereto. The original mining source data corresponding to the target mining source may refer to mining source data obtained when the target mining source is about to perform big data acquisition and last data acquisition before mining.

Step S130, a target mining network is called, and a target mining result corresponding to the target mining source is obtained based on the original mining source data and the updated user behavior data.

In this embodiment, the target mining result may be used to indicate a target attribute of each mining object in the target mining source, and the target attribute of any mining object may be used to indicate that any mining object belongs to the target big data mining policy or that any mining object does not belong to the target big data mining policy.

And step S140, updating the target big data mining strategy in the target mining source based on the target mining result to obtain a global updating result of the target big data mining strategy in the target mining source, and mining the big data based on the global updating result.

In this embodiment, after the global update result of the target big data mining strategy in the target mining source is obtained, the updated target big data mining strategy can be obtained, and therefore, the big data mining is performed based on the updated target big data mining strategy, and a more accurate big data mining result can be obtained.

Based on the above steps, in this embodiment, a target mining result corresponding to the target mining source is automatically obtained based on the original mining source data and the updated user behavior data corresponding to the target mining source, and then a global update result of the target big data mining strategy in the target mining source is automatically obtained based on the target mining result. Based on the process, the big data mining strategy can be automatically updated, the updating process of the big data mining strategy does not depend on manpower, the efficiency of updating the big data mining strategy in the mining source is improved, and the reliability of the updating result of the obtained big data mining strategy is high.

In one possible implementation manner, for step S130, in the process of obtaining a target mining result corresponding to the target mining source based on the original mining source data and the updated user behavior data for invoking the target mining network, the following exemplary sub-steps may be implemented, which are described in detail below.

And the substep S131, calling a target mining network, and sequentially executing forward mining analysis of a first iteration strategy based on the original mining source data and the associated characteristic data of the updated user behavior data to obtain a first mining component corresponding to the target mining source.

In this embodiment, the associated feature data may refer to feature data obtained by fusing updated user behavior data with original mining source data, and the forward mining analysis may refer to mining analysis for some forward behaviors, for example, a behavior of a user concerning a certain feature object may be understood as a forward behavior.

And a substep S132, sequentially executing negative excavation analysis of the first iteration strategy based on the target distinguishing excavation characteristics corresponding to the first excavation component to obtain a second excavation component corresponding to the target excavation source.

In this embodiment, the target-specific mining feature corresponding to the first mining component may be a mining feature obtained by performing target-specific mining on all relevant mining components of the service unit correspondingly covered by the first mining component, and the negative mining analysis may be mining analysis for some negative behaviors, for example, a behavior of a user cancelling attention of a certain feature object may be understood as a negative behavior.

And a substep S133 of performing target distinguishing mining analysis on the second mining component to obtain a target mining result corresponding to the target mining source.

In this embodiment, the differentiated mining analysis may refer to mining analysis performed in different dimensions with respect to the mining component, for example, the mining analysis may be performed from dimensions of different business boards, and is not limited specifically.

In a possible implementation, the first iterative strategy may include a preset number of times of iterative mining, for example, the preset number of times may be three times, and any one positive mining analysis includes one differential mining analysis and one frequent pattern mining analysis. The frequent pattern mining analysis is to compress the distinguishing mining features generating the frequent sets into a frequent pattern tree FP-tree, store the associated information of the entries by the FP-tree, and then generate the frequent sets for the pattern tree to obtain further distinguishing mining features.

Thus, for sub-step S131, it can be realized by the following exemplary embodiments.

(1) And performing first distinguishing mining analysis on the associated feature data of the original mining source data and the updated user behavior data to obtain a first distinguishing mining feature corresponding to the target mining source.

(2) And carrying out first frequent pattern mining analysis on the first distinguishing mining characteristics to obtain first frequent pattern mining characteristics corresponding to the target mining source.

(3) And carrying out second distinguishing mining analysis on the first frequent pattern mining characteristics to obtain second distinguishing mining characteristics corresponding to the target mining source.

(4) And performing second frequent pattern mining analysis on the second distinguishing mining characteristics to obtain second frequent pattern mining characteristics corresponding to the target mining source.

(5) And carrying out third region mining analysis on the second frequent pattern mining characteristics to obtain third region mining characteristics corresponding to the target mining source.

(6) And carrying out third frequent pattern mining analysis on the third region mining characteristics to obtain a first mining component corresponding to the target mining source.

Further, in one possible implementation, any of the negative mining analyses described above may include an inverse discriminative mining analysis and a discriminative mining analysis. The inverse differential mining analysis is a processing mode aiming at the inverse dimension of the differential mining analysis.

Thus, for sub-step S132, it can be realized by the following exemplary embodiments.

(1) And carrying out first inverse differential mining analysis on the target differential mining characteristics corresponding to the first mining component to obtain first negative mining characteristics corresponding to the target mining source.

(2) And performing fourth distinguishing mining analysis on the splicing characteristics of the first negative mining characteristic and the third distinguishing mining characteristic to obtain a fourth distinguishing mining characteristic corresponding to the target mining source.

(3) And carrying out second inverse differential excavation analysis on the fourth differential excavation characteristics to obtain second negative excavation characteristics corresponding to the target excavation source.

(4) And performing fifth distinguishing mining analysis on the second negative mining characteristics and the splicing characteristics of the second distinguishing mining characteristics to obtain fifth distinguishing mining characteristics corresponding to the target mining source.

(5) And performing third inverse differential excavation analysis on the fifth differential excavation feature to obtain a third negative excavation feature corresponding to the target excavation source.

(6) And carrying out sixth distinguishing mining analysis on the third negative mining characteristics and the splicing characteristics of the first distinguishing mining characteristics to obtain a second mining component corresponding to the target mining source.

Further, in order to facilitate evaluating reliability of the global update result obtained based on the target mining result, after step S130, the method provided in the embodiment of the present application may further include step S134 of invoking a target correction network, and obtaining target correction reference degree information based on the original mining source data and the target mining result.

The target correction reference degree information is used for indicating the reliability of the global updating result obtained based on the target mining result. Illustratively, the target correction reference degree information includes a reference degree of which the target mining result is a correct mining result and a reference degree of which the target mining result is an incorrect mining result, and the sum of the reference degree of which the target mining result is a correct mining result and the reference degree of which the target mining result is an incorrect mining result is 1. If the reference degree of the target mining result which is the correct mining is not less than the reference degree of the target mining result which is the wrong mining result, the reliability of the global updating result obtained based on the target mining result is considered to be higher; if the reference degree of the target mining result which is correct mining is smaller than the reference degree of the target mining result which is wrong mining result, the reliability of the global updating result determined based on the target mining result is considered to be lower, and at the moment, the global updating result of the target tree-shaped organization determined based on the target mining result in the target image is possibly wrong and needs to be corrected manually.

In a possible implementation manner, the target correcting network may include at least one distinguishing mining network element, at least one merging network element and one correcting classification network element which are connected in sequence.

Thus, step S134 can be implemented by the following exemplary embodiments.

(1) And inputting the original mining source data and the target mining result into a first differentiated mining network unit in the target correction network for processing to obtain differentiated mining characteristics output by the first differentiated mining network unit.

(2) And from the second differentiated mining network unit, inputting the differentiated mining features output by the previous differentiated mining network unit into the next differentiated mining network unit for processing to obtain the differentiated mining features output by the next differentiated mining network unit.

(3) And inputting the distinguishing mining characteristics output by the last distinguishing mining network unit into the first converged network unit for processing to obtain the converged characteristics output by the first converged network unit.

(4) And from the second converged network unit, inputting the converged features output by the previous converged network unit into the next converged network unit for processing to obtain the converged features output by the next converged network unit.

(5) And inputting the fusion characteristics output by the last fusion network unit into the correction classification network unit for processing to obtain target correction reference degree information output by the correction classification network unit.

In one possible implementation, the configuration process of the target correcting network and the target mining network is exemplarily described below in connection with one possible example, and may be specifically implemented by the following exemplary embodiments.

(1) The method comprises the steps of obtaining original tag mining source data corresponding to at least one tag mining source, updated reference tag data corresponding to at least one tag mining source and standard mining results corresponding to at least one tag mining source.

(2) And calling an initial mining network, and acquiring a prediction mining result corresponding to a first label mining source based on original label mining source data corresponding to the first label mining source in at least one label mining source and updated reference label data corresponding to the first label mining source.

(3) And calling an initial correction network, and acquiring first correction reference degree information based on original tag mining source data corresponding to the first tag mining source and a prediction mining result corresponding to the first tag mining source.

(4) And acquiring second correction reference degree information based on the original tag mining source data corresponding to the first tag mining source and the standard mining result corresponding to the first tag mining source.

(5) The first difference information is determined based on the first correction reference information and the second correction reference information.

(6) Parameters of the initial correction network are updated based on the first difference information.

(7) And responding to the updating process of the parameters of the initial correction network to meet the first training end condition, and obtaining a first correction network.

(8) And calling an initial mining network, and acquiring a prediction mining result corresponding to a second label mining source based on original label mining source data corresponding to the second label mining source in at least one first label mining source and updated reference label data corresponding to the second label mining source.

(9) And calling a first correction network, and acquiring third correction reference degree information based on the original tag mining source data corresponding to the second tag mining source and the prediction mining result corresponding to the second tag mining source.

(10) And determining second difference information based on the third correction reference degree information, the predicted mining result corresponding to the second label mining source and the standard mining result corresponding to the second label mining source.

(11) And updating the parameters of the initial mining network based on the second difference information.

(12) And responding to the updating process of the parameters of the initial mining network to meet a second training end condition, and obtaining the first mining network.

(13) And in response to the condition that the constraint training process does not meet the target training end condition, continuing to perform constraint training on the first correction network and the first mining network until the constraint training process meets the target training end condition, and obtaining a target correction network and a target mining network.

For example, the constraint training may be a method of generating a training against the constraint training, and is not particularly limited.

In a possible implementation manner, the target mining source is an initial at least partial mining source corresponding to the target big data mining strategy in the initial mining source. Thus, after step S140, in order to obtain a more accurate global update result. The following steps, described in detail below, may also be included.

And S150, responding to the situation that the global updating result of the target big data mining strategy in the target mining source does not meet the updating training end condition, and acquiring a next at least partial mining source corresponding to the target big data mining strategy in the initial mining source based on the global updating result of the target big data mining strategy in the target mining source.

And step S160, acquiring a global updating result of the target big data mining strategy in the next at least partial mining source.

Step S170, responding to that the global updating result of the target big data mining strategy in the next at least partial mining source meets the updating training end condition, and acquiring the global updating result of the target big data mining strategy in the initial mining source based on the acquired global updating result of the target big data mining strategy in each at least partial mining source.

For example, in another possible implementation, the target mining network described above may also be obtained by:

(2) And performing supervision training on the initial mining network based on the original tag mining source data corresponding to the at least one tag mining source, the updated reference tag data corresponding to the at least one tag mining source and the standard mining result corresponding to the at least one tag mining source respectively to obtain the target mining network.

In a possible implementation manner, the target mining source may be obtained from a service subscription mining source including a target mining node corresponding to the target big data mining policy. It should be noted that the target attribute of any mining object may be used to indicate that any mining object belongs to the target mining node or that any mining object does not belong to the target mining node.

In this way, for step S140, in the process of updating the target big data mining policy in the target mining source based on the target mining result to obtain the global update result of the target big data mining policy in the target mining source, the following exemplary sub-steps may be implemented, which are described in detail below.

In the substep S141, a target excavation object belonging to the target excavation node is determined among the excavation objects in the target excavation source based on the target excavation result.

And the substep S142, based on the target mining object, marking the mining service segments of the target mining nodes and the incidence relation between the mining service segments of the target mining nodes in the target mining source to obtain a target marking result.

And a substep S143, acquiring a global updating result of the target mining node in the target mining source based on the target marking result.

In this embodiment, for example, the mining service segments with the association relationship may be expanded, added, or updated to obtain a global update result of the target mining node in the target mining source.

In a possible implementation manner, further to step S110, in the process of obtaining an acquisition configuration result of the acquisition configuration output for the acquisition feature item, the following exemplary sub-steps may be implemented, which are described in detail below.

And a substep S111, acquiring a service functional area acquisition process obtained from the current information push source based on the information push strategy.

And a substep S112, acquiring the acquired big data label information of a plurality of key operation acquisition nodes in the key acquisition node operation distribution of the service functional area acquisition process.

In this embodiment, each key operation acquisition node may be configured to represent one or more acquisition page objects that need to be acquired in an acquisition preparation process of the service functional area, and an acquisition feature item of an acquisition page object represented by each key operation acquisition node needs to be acquired and activated.

In this embodiment, the collected page object may refer to a specific page function element (e.g., live attention, commodity collection, etc.) of a collected data page (e.g., a browsing page of an e-commerce live broadcast).

In this embodiment, the collected big data tag information of any key operation collecting node is used to reflect the collection type relationship between any key operation collecting node and other key operation collecting nodes. For example, for e-commerce live broadcast service, a key operation acquisition node of live broadcast attention service, a key operation acquisition node of live broadcast ordering service, and a key operation acquisition node of live broadcast recommendation service have an acquisition type relationship, that is, after live broadcast attention, live broadcast ordering may be performed, and after live broadcast ordering, live broadcast recommendation of live broadcast goods may be performed.

And a substep S113, layering at least two key operation acquisition node services into a target service layered acquisition plan according to the acquired big data label information of each key operation acquisition node.

In this implementation, the target service hierarchical acquisition plan is used to instruct acquisition configuration of acquisition feature items of acquisition page objects represented by the key operation acquisition nodes of the service hierarchy. The service layering may refer to hierarchical clustering division performed on actually acquired services of the key operation acquisition nodes.

And a substep S114, updating the running distribution of the key acquisition nodes by adopting the target service hierarchical acquisition plan, and sending the updated running distribution of the key acquisition nodes to the acquisition configuration process of the software acquisition plan.

In this implementation, the updated key acquisition node operation distribution may be used to instruct the acquisition configuration process of the software acquisition plan to perform acquisition configuration on the acquisition feature items of the acquisition page object represented by the key operation acquisition nodes of the service hierarchy in the acquisition preparation process of the service functional area acquisition plan according to the instruction of the target service hierarchical acquisition plan, and output the acquisition configuration result.

In detail, in some possible implementation manners, the collection feature item may be configured according to a requirement of an actual software function, and may specifically be customized, or refer to a conventional collection configuration task in the prior art, which is not limited herein. In addition, the specific acquisition configuration logic of the acquisition configuration process of the software acquisition plan may perform adaptive configuration with reference to the acquisition feature items, and the content and form of the specific acquisition configuration are not the technical problems that the embodiments of the present application aim to solve, and will not be described in detail herein.

Based on the above steps, in this embodiment, at least two key operation collection node services may be layered into a target service layered collection plan according to the collected big data tag information of a plurality of key operation collection nodes in the key operation collection node operation distribution of the service functional area collection process, where the target service layered collection plan is used to instruct to collect and configure the collection feature items of the collection page object represented by the key operation collection nodes of the service layered. Then, the operation distribution of the key acquisition nodes can be updated by adopting the target business layered acquisition plan, and the updated operation distribution of the key acquisition nodes is sent to the acquisition configuration process of the software acquisition plan, so that the acquisition configuration process of the software acquisition plan can acquire and configure the acquisition characteristic items of the acquisition page objects represented by the key operation acquisition nodes of the business layers according to the indication of the target business layered acquisition plan in the process of acquiring and configuring the acquisition process of the business functional area, thereby reducing the times of re-calling the acquisition configuration, saving the induction time of the acquisition data and improving the execution efficiency of the acquisition plan.

In a possible implementation manner, before the embodiments of the present application are described in detail, the following first explains the respective defined terms so that those skilled in the art can clearly and completely realize the scheme of the embodiments of the present application.

In detail, the plurality of key operation collection nodes correspond to a target information collection chain. The target information acquisition chain is obtained by connecting a plurality of acquisition and migration chains with each key operation acquisition node based on the acquisition chain distribution relation of each key operation acquisition node in the key acquisition node operation distribution.

As such, in the art, the acquisition chain distribution relationship may be used to indicate: and one key operation acquisition node is matched with the acquisition relation of other key operation acquisition nodes along at least one acquisition operation chain in the key acquisition node operation distribution.

The collection plan data label information of any key operation collection node comprises at least one of the following items: an acquisition plan partition sequence for any key operational acquisition node and an inverse acquisition plan partition sequence for any key operational acquisition node.

It is worth to be noted that the acquisition plan partition in the acquisition plan partition sequence of any critical operation acquisition node may be understood as: and the key operation acquisition nodes are covered by all forward acquisition modes from the first key operation acquisition node to any key operation acquisition node in the target information acquisition chain.

The collection plan partition sequence of any key operation collection node is separated from the collection plan partition with the first key operation collection node and is the first collection plan partition of any key operation collection node.

In this embodiment, the inverse acquisition plan partition in the inverse acquisition plan partition sequence of any key operation acquisition node may be understood as: and the key operation acquisition nodes are covered by all reverse acquisition modes from the first key operation acquisition node to any key operation acquisition node in the reverse acquisition relation corresponding to the target information acquisition chain.

In this embodiment, the inverse acquisition plan partition sequence of any key operation acquisition node is separated from the first inverse acquisition plan partition of any key operation acquisition node, and is the first inverse acquisition plan partition of any key operation acquisition node.

In this embodiment, the inverse acquisition relationship refers to an acquisition relationship obtained by performing inverse processing on each acquisition migration chain in the target information acquisition chain.

In this way, for step S113, in the process of layering at least two key operation collection node services into a target service layered collection plan according to the collected big data tag information of each key operation collection node, the following exemplary substeps may be implemented, which are described in detail below.

And a substep S1131, constructing an acquisition plan data label network formed by a plurality of key operation acquisition nodes according to the acquired big data label information of each key operation acquisition node.

And a substep S1132 of extracting service hierarchical distribution information based on the acquisition plan data tag network.

In this embodiment, the service hierarchical distribution information may include: and acquiring plan sequences required by the multi-layer service layering, wherein at least one acquiring plan is a key operation acquiring node in each acquiring plan sequence.

And a substep S1133 of performing at least one layer of service hierarchical iterative processing on the plurality of key operation acquisition nodes according to the service hierarchical distribution information to obtain a target service hierarchical acquisition plan.

For example, in a possible implementation manner, a w acquisition plan sequence required by a w-th service hierarchy may be determined according to service hierarchy distribution information, and a total acquisition plan hop count of the w-th acquisition plan sequence may be determined according to an acquisition plan hop count of each acquisition plan in the w-th acquisition plan sequence; and W belongs to [1, W ], wherein W is the distribution layer level number of the service hierarchical distribution information. And when the total number of acquisition plan jumping times of the w-th acquisition plan sequence is less than or equal to the acquisition plan jumping time threshold, performing service layering processing on each acquisition plan in the w-th acquisition plan sequence to obtain a w-th service layering acquisition plan. And if the current value of W is less than W and the total acquisition plan jump times of the W +1 acquisition plan sequence required by the W +1 th service hierarchy acquired according to the service hierarchy level information is greater than the acquisition plan jump time threshold, acquiring a target service hierarchy acquisition plan according to the W-th service hierarchy acquisition plan.

In one possible implementation, the sub-step S1131 can be implemented by the following embodiments.

(1) And taking the first key operation acquisition node in the target information acquisition chain as a reference acquisition plan of an acquisition plan data label network, and determining the rest key operation acquisition nodes except the first key operation acquisition node in the target information acquisition chain in the plurality of key operation acquisition nodes.

(2) And acquiring the first acquisition plan partition of each remaining key operation acquisition node from the acquisition plan partition sequence in the acquired big data label information of each remaining key operation acquisition node.

(3) And determining the first acquisition type relationship among the key operation acquisition nodes according to the first acquisition plan partition of each remaining key operation acquisition node.

(4) And adding each remaining key operation acquisition node to the reference acquisition plan according to the relationship of the first acquisition type so as to obtain an acquisition plan data label network.

In a possible implementation manner, the parent acquisition plan of each key operation acquisition node in the acquisition plan data tag network except for the reference acquisition plan is as follows: the first acquisition plan partition for each critical operation acquisition node. W collection plan data label combinations exist in a plurality of key operation collection nodes, and one collection plan data label combination is associated with a collection plan sequence required by at least one business layer. Wherein W is a positive integer.

Thus, the sub-step S1132 can be implemented by the following embodiments.

(1) And selecting a first key operation acquisition node from key operation acquisition nodes which are not subjected to the targeted processing in the acquisition plan data label network according to the targeted processing sequence of the label priority.

(2) And detecting whether a W-th collection plan data label combination formed by a second key operation collection node and a first key operation collection node exists or not according to the inverse collection plan partition sequence of each key operation collection node except the last key operation collection node in the target information collection chain, wherein W belongs to [1, W ].

(3) The second key operation acquisition node meets the following conditions: the second critical operation collection node is the first collection plan partition of the first critical operation collection node, and the first critical operation collection node is the first inverse collection plan partition of the second critical operation collection node.

(4) If the target service hierarchy exists, selecting at least one key operation acquisition node from a plurality of key operation acquisition nodes according to a second key operation acquisition node, adding the at least one key operation acquisition node into an acquisition plan sequence required by the target service hierarchy associated with the w-th acquisition plan data label combination, and continuing to process the acquisition plan data label network in a targeted manner. And if not, reselecting the first key operation acquisition node until all key operation acquisition nodes in the acquisition plan data label network are processed in a targeted manner.

For example, if the acquisition plan data tag exists, an extended acquisition plan sequence of a second key operation acquisition node may be acquired from the acquisition plan data tag network, and if the extended acquisition plan sequence only includes a first key operation acquisition node and an extended acquisition plan of the first key operation acquisition node, the first key operation acquisition node and the second key operation acquisition node are selected and added to an acquisition plan sequence required by a target service hierarchy associated with a w-th acquisition plan data tag combination. Or if the extended acquisition plan sequence comprises other extended acquisition plans except the first key operation acquisition node and the extended acquisition plan of the first key operation acquisition node, selecting other extended acquisition plans to add to the acquisition plan sequence required by the target service hierarchy.

Illustratively, in the process of selecting a first key operation acquisition node and a second key operation acquisition node and adding the first key operation acquisition node and the second key operation acquisition node to an acquisition plan sequence required by a target service hierarchy associated with a w-th acquisition plan data tag combination, it may be detected whether a first historical acquisition plan sequence including the first key operation acquisition node exists in an acquisition plan sequence required by a historical layer service hierarchy associated with the previous w-1 acquisition plan data tag combinations.

For example, if there is a first historical acquisition plan sequence, adding a service hierarchical acquisition plan corresponding to the first historical acquisition plan sequence and a second key operation acquisition node to an acquisition plan sequence required by a target service hierarchy associated with a w-th acquisition plan data tag combination.

For another example, if there is no first historical acquisition plan sequence, then a first critical operation acquisition node and a second critical operation acquisition node are added to the acquisition plan sequence required by the target business hierarchy.

In a possible implementation manner, in the process of selecting other extended acquisition plans to add to the acquisition plan sequence required by the target service hierarchy, it may be specifically detected whether a second historical acquisition plan sequence exists in the acquisition plan sequences required by the historical layer service hierarchies associated with the previous w-1 acquisition plan data tag combinations, where the second historical acquisition plan sequence includes the service hierarchy acquisition plans corresponding to the other extended acquisition plans.

For example, if a second historical acquisition plan sequence exists, a service hierarchical acquisition plan, a first key operation acquisition node and a second key operation acquisition node corresponding to the second historical acquisition plan sequence are added to an acquisition plan sequence required by a target service hierarchy.

For another example, if there is no second historical acquisition plan sequence, then add the other extended acquisition plan to the acquisition plan sequence required for the target business hierarchy, and add the business hierarchy acquisition plan, the first critical operation acquisition node, and the second critical operation acquisition node of the other extended acquisition plan business hierarchy to the acquisition plan sequence required for the next tag business hierarchy below the target business hierarchy associated with the w-th acquisition plan data tag combination.

In one possible implementation manner, for step S114, in the process of updating the operation distribution of the key acquisition nodes by using the target business hierarchical acquisition plan, the following exemplary sub-steps may be implemented, which are described in detail below.

And a substep S1141 of adding a target service layered acquisition plan in the operation distribution of the key acquisition nodes, and connecting the target service layered acquisition plan and the service layered key operation acquisition nodes by adopting an acquisition and migration chain.

And a substep S1142 of adding a matched acquisition plan for the key operation acquisition nodes which are not layered by the service in the operation distribution of the key acquisition nodes, and adding a matched acquisition plan for the layered acquisition plan of the target service in the operation distribution of the key acquisition nodes.

In a possible implementation manner, for step S111, in the process of acquiring the service functional area acquisition process obtained based on the current information push source of the information push policy, the following exemplary sub-steps may be implemented, which are described in detail below.

In sub-step S1111, the information push operation container of the digital content subscription device 200 is obtained, and the information push policy analysis is performed on the information push operation container through the information push service, so as to obtain the information push policy information of the information push source in the information push operation container.

In the substep S1112, a service functional region is analyzed based on the information push policy information of the information push source to obtain a target service functional region of the information push source.

And a substep S1113, analyzing the collected time-space sequence control record information of the information push operation container based on the collected time-space sequence control script to obtain the collected time-space sequence control record information of the information push source.

Substep S1114 performs service functional area update on the target service functional area of the information push source and the acquisition time-sequence control record information of the information push source in the information push operation container to obtain service functional area update information of the information push source, and performs information acquisition partition positioning on the information push policy based on the service functional area update information of the information push source to obtain a current information acquisition policy of the information push source.

In this embodiment, the information push service may be understood as an information push program running in the cloud, for example, each information push module in the information push process may be configured in the cloud in advance, for example, a function module that needs to use information push policy analysis in this embodiment, and then this operation is performed. The operation of the function module related to the specific information push policy resolution can be referred to the following detailed description of step S1111.

In this embodiment, the information push operation container may be understood as a cloud computing container formed by information push policies bound to information push sources that are generated based on the user feedback representation. The information push policy may refer to a policy package generated by a large amount of big data when the digital content subscription apparatus 200 performs any information push software. The information push source may refer to a certain data source formed under the information push policy.

In this embodiment, the service function area may be used to represent service function logic distinguishing information of a service function corresponding to each information push source, and the acquisition time-space sequence control record information may be used to describe the service function logic distinguishing information for each acquisition unit in the acquisition time-space sequence control process.

In the information pushing process, the service function area needs to be considered, so that the operation of targeted acquisition configuration is performed according to the feature distribution of the service function area. In this way, in the embodiment, the target service functional area of the information push source and the acquisition time-space sequence control record information of the information push source are updated through the service functional area, so that the information push strategy information of the information push source and the acquisition time-space sequence control record information of the information push source are integrated, and the service functional area of the rich information push source is extracted, thereby facilitating the efficient acquisition time-space sequence control based on the service functional area of the information acquisition strategy in the information acquisition strategy updating process; in addition, the information acquisition partition positioning is carried out on the information push source through the target service functional area of the information push source to obtain the current information acquisition strategy of the information push source, so that the information acquisition strategy of the modular unit is convenient to update, and the update restart time of the information acquisition strategy after the information acquisition strategy fails to update is reduced.

In a possible implementation manner, for step S1112, in the process of acquiring collected big data tag information of a plurality of key operation collection nodes in the operation distribution of key collection nodes of the service functional area collection process, each key collection node in the service functional area collection process may be extracted to construct the operation distribution of key collection nodes according to the service relationship of each key collection node, and the collected big data tag information of each key collection node is acquired according to the collection type relationship between each key operation collection node and other key operation collection nodes.

It should be noted that the collection type relationship between each key operation collection node and other key operation collection nodes is obtained from the process configuration information in the collection process of the service functional area.

While certain alternative embodiments of the present application will be described below with reference to the above, it should be understood that the following description of the embodiments is only exemplary and should not be taken as an exhaustive identification of the features necessary to implement the present invention.

In a possible implementation manner, for step S1111, in the process of performing information push policy analysis on the information push operation container through the information push service to obtain the information push policy information of the information push source in the information push operation container, the following exemplary sub-steps may be implemented, which are described in detail below.

In the substep S11111, a policy container logical pointer data set bound by the push logical pointer controller of each push source in the information push operation container is obtained.

In this embodiment, it is worth to be noted that the policy container logical pointer data set includes policy container logical pointer data using each logical pointer rule as a reference unit, and the policy container logical pointer data includes a logical pointer data source condition of the logical pointer rule, a logical pointer data source result, and an index operation node of a logical pointer data source in the logical pointer rule. For example, the logical pointer rule may be used to represent a business scope related to a code editing configuration process, the logical pointer data source condition may be used to characterize a condition of the logical pointer data source (e.g., a condition may be used when a certain function code is called), and the logical pointer data source result may be used to characterize a running result indicated after the logical pointer data source.

For example, in the present embodiment, the push logical pointer controller may be a software program, and the push logical pointer controller refers to a program having a function of pushing logical pointer data.

Substep S11112, for each logical pointer rule, according to each index operation node row of a plurality of index operation node rows in the index operation node of the logical pointer data source of the logical pointer rule from each push source, determining whether each stored structure interpretation information in the index operation node row is a new information acquisition policy update code acquisition plan according to the interpretation vector representation of the stored structure interpretation information in the index operation node row, according to the acquisition plan type of the new information acquisition policy update code acquisition plan in the index operation node row, determining a logical pointer data page of each code resource packet corresponding to the index operation node row, for the logical pointer data page of each code resource packet, dividing the logical pointer data page of the code resource packet into a plurality of sub logical pointer data pages, according to the interpretation tag and the preset storage interpretation range of each stored structure interpretation information in each sub logical pointer data page, it is determined whether the logical pointer data page of the code resource package is a logical pointer data page of the target logical pointer package.

It is worth mentioning that each storage structure interpretation information updates the classified storage structure behavior corresponding to each information collection policy.

And a substep S11113, obtaining information acquisition strategy update classification template block information of each storage structure interpretation information in the logic pointer data page of the preset information acquisition strategy update classification template matching target logic pointer program package, wherein the information acquisition strategy update classification template block information comprises an index main key function label of the information acquisition strategy and an index main key application label of the information acquisition strategy, and the preset information acquisition strategy update classification template comprises matching strategies corresponding to index main key use types of different information acquisition strategies.

Substep S11114, determining the variable information of the update cycle variable of the information acquisition strategy of the index main key of each information acquisition strategy and the constant information of the update cycle constant of each information acquisition strategy according to the information acquisition strategy update classification template block information of the index operation node of each logic pointer data source of each different logic pointer rule in the strategy container logic pointer data set, determining the index main key tag combination object of the information acquisition strategy of each push source in the logic pointer rule according to the variable information of the update cycle variable of the information acquisition strategy of the index main key of each information acquisition strategy and the constant information of the update cycle constant of each information acquisition strategy in the logic pointer data page of the target logic pointer program package, and selecting the index ordering information in the index selection object range of the index main key of the information acquisition strategy of the index main key tag combination object of the information acquisition strategy and the index main key tag combination object of the information acquisition strategy And after the index sorting information of the index selection object range of the index main key of the information acquisition strategy of the label combination object is used as the acquisition time-space sequence control script information of each push source in the logic pointer rule, the acquisition time-space sequence control script information of each push source in all the logic pointer rules is gathered to obtain the information push strategy information of the information push sources in the information push operation container.

For example, an update cycle constant of an information collection policy refers to data whose value remains unchanged throughout the operation, and is usually given directly in a command or program, and the types of data used as the constant are numeric, character, date, logical, and monetary. For another example, the update cycle variable of the information collection policy refers to data of the value policy in the whole operation process, and has an abstraction of a storage space, and the update cycle variable of the information collection policy is a placeholder convenient to use and used for referencing a memory address, and the address can store modifiable program information during Script operation.

In a possible implementation manner, for step S1112, in the process of performing service functional area analysis based on the information push policy information of the information push source to obtain the target service functional area of the information push source, the following exemplary sub-steps may be implemented, which are described in detail below.

And a substep S11121 of obtaining primary key constraint index object allocation information of a primary key constraint index object set added to the logical pointer data source description information of each acquired spatio-temporal sequence control script information in the information push strategy information of the information push source, and determining a first primary key constraint key word set corresponding to the primary key constraint index object allocation information.

For example, there may only be one aggregated index in a table, but each column in the table may have its own non-aggregated index. If a primary key constraint is created in the table, SQL Server will automatically generate a uniqueness constraint for it. When creating primary key constraints, if a CLUSTERED key is formulated, a unique aggregate index will be generated for the table.

It should be noted that the primary key constraint index object allocation information includes relationship graph collection plan object information for calculating relationship information according to relationship graph rules determined by relationship graph rule input information and relationship graph rule output information of the primary key constraint index object set, and the first primary key constraint key set includes a high-low order of priorities of a plurality of relationship graph collection plans of the relationship graph collection plan object information.

And a substep S11122, determining a first relation graph rule characteristic of the logic pointer data source description information of each acquisition time-space sequence control script information based on the relation graph rule input information and a second relation graph rule characteristic based on the relation graph rule output information.

And a substep S11123, determining a constraint business function positioning parameter for performing constraint business function positioning on the first main key constraint keyword set according to the collection plan priority relationship of the relation graphs of the first relation graph rule characteristic and the second relation graph rule characteristic.

And a substep S11124, performing constraint business function positioning on the first primary key constraint keyword set based on the constraint business function positioning parameter to obtain a second primary key constraint keyword set.

And a substep S11125, performing relationship graph type division on the second primary key constraint keyword set to obtain a plurality of relationship graph type division sets, and performing feature extraction on each relationship graph type division set to obtain a relationship graph type division variable.

And a substep S11126, which is to determine the service function region for each collected time-space sequence control script information according to the service function regions corresponding to the plurality of relation graph type division variables corresponding to the second primary key constraint keyword set.

And a substep S11127 of obtaining a target service functional area of the information push source based on each service functional area of the collected time-sequence control script information.

Further, in a possible implementation manner, for step S1113, in the process of analyzing the acquisition-spatio-temporal sequence control record information of the information push operation container based on the acquisition-spatio-temporal sequence control script to obtain the acquisition-spatio-temporal sequence control record information of the information push source, the following exemplary sub-steps may be implemented, which are described in detail as follows.

And a substep S11131, inputting the information pushing operation container into a preset acquisition time-space sequence control script to obtain the matching degree of the information pushing operation container matched with the update rule of each preset information acquisition strategy.

And a substep S11132, determining the update rule of the target information collection policy corresponding to the information push operation container according to the matching degree of the information push operation container and the update rule of each preset information collection policy.

For example, the update rule of the preset information collection policy with the matching degree greater than the preset matching degree threshold may be determined as the update rule of the target information collection policy corresponding to the information pushing operation container.

And a substep S11133 of extracting acquisition time-sequence control record information matched with each information push source from the update rule description information of the information acquisition policy of the update rule of the target information acquisition policy corresponding to the information push operation container.

For example, in the extraction process, the feature information with the structured description information, which is matched with each information push source, in the update rule description information of the information collection policy may be specifically extracted.

In a possible implementation manner, for step S1114, in the process of performing service function area update on the target service function area of the information push source in the information push operation container and the collected time-sequence control record information of the information push source to obtain the service function area update information of the information push source, the following exemplary sub-steps may be implemented, which are described in detail below.

And a substep S11141, adding the target service functional area of the information push source and the collected time-space sequence control record information into a service functional area updating program, and determining a relation diagram knowledge diagram of the target service functional area of the information push source and the collected time-space sequence control record information corresponding to the updating base table of each information collection strategy from the service functional area updating program.

And a substep S11142, combining the knowledge graphs of the relationship graphs for carrying out relationship graph type division according to the relationship labels of the base table relationship between the updated base tables of the information acquisition strategies in the knowledge graphs of the different relationship graphs to obtain at least one knowledge graph sequence of the relationship graphs.

In this embodiment, the relationship labels of the base table relationship of the updated base table of the information acquisition strategies in any two relationship map knowledge graphs in the same relationship map knowledge graph sequence cover the preset storage interpretation range.

And a substep S11143, determining, for each relation map knowledge map sequence, program relation map call information corresponding to the relation map knowledge map sequence for the target service function area of the information push source and the collected time-sequence control record information from the service function area update program based on each relation map knowledge map in the relation map knowledge map sequence.

In this embodiment, the program relation diagram call information at least includes characteristic information data of each relation diagram distribution of each relation diagram knowledge diagram in the relation diagram knowledge diagram sequence for a target service function region of an information push source and collection time-sequence control record information, and the program relation diagram call information is used for determining a relation distribution result of a relation diagram relation between the target service function region of the information push source and collection time-sequence control record information corresponding to an update base table of an information collection strategy in each relation diagram knowledge diagram in the relation diagram knowledge diagram sequence.

And a substep S11144, determining a relation distribution result of the relation diagram relation corresponding to the update base table of the information acquisition strategy in each relation diagram knowledge diagram in the relation diagram knowledge diagram sequence based on the program relation diagram calling information, classifying the relation distribution result of the relation diagram relation by adopting a preset base table relation correction network to obtain a classification result, and obtaining the service function area update information of the information push source according to the classification result.

In a possible implementation manner, still referring to step S1114, in the process of performing information acquisition partition location on the information push policy based on the service functional area update information of the information push source to obtain the current information acquisition policy of the information push source, the following exemplary sub-steps may be implemented, which are described in detail below.

And a substep S11145, obtaining the positioning information of the information acquisition partition of the information push source under the information push strategy.

In this embodiment, the information collection partition may refer to a logical partition of each information collection data area under the information pushing policy, for example, a service logical partition, a functional partition where a user subscribes to a service, and the like.

And a substep S11146, obtaining the information acquisition partition positioning items under the positioning information of the information acquisition partition and the base table relation configuration information corresponding to each information acquisition partition positioning item.

And a substep S11147, which covers and configures the service functional area update information of the information push source under the base table relationship configuration information corresponding to each information acquisition partition positioning item, so as to obtain the current information acquisition strategy of the information push source.

Fig. 3 is a schematic diagram of functional modules of a cloud computing and artificial intelligence based big data mining apparatus 300 according to an embodiment of the present disclosure, in this embodiment, the cloud computing and artificial intelligence based big data mining apparatus 300 may be divided into the functional modules according to the method embodiment executed by the digital content center 100, that is, the following functional modules corresponding to the cloud computing and artificial intelligence based big data mining apparatus 300 may be used to execute the method embodiments executed by the digital content center 100. The cloud computing and artificial intelligence based big data mining device 300 may include a first obtaining module 310, a second obtaining module 320, a calling module 330, and an updating module 340, and the functions of the functional modules of the cloud computing and artificial intelligence based big data mining device 300 are described in detail below.

The first obtaining module 310 is configured to obtain a collection configuration result that is output by performing collection configuration on collection feature items, where the collection configuration result includes a target big data mining policy for each collection feature item. The first obtaining module 310 may be configured to perform the step S110, and for a detailed implementation of the first obtaining module 310, reference may be made to the detailed description of the step S110.

A second obtaining module 320, configured to obtain a target mining source corresponding to the target big data mining policy, original mining source data corresponding to the target mining source, and updated user behavior data corresponding to the target mining source, where the updated user behavior data is determined and obtained based on at least a partial update result of the target big data mining policy in the target mining source. The second obtaining module 320 may be configured to perform the step S120, and for a detailed implementation of the second obtaining module 320, reference may be made to the detailed description of the step S120.

The invoking module 330 is configured to invoke a target mining network, and obtain a target mining result corresponding to the target mining source based on the original mining source data and the updated user behavior data, where the target mining result is used to indicate a target attribute of each mining object in the target mining source, and a target attribute of any mining object is used to indicate that the any mining object belongs to the target big data mining policy or that the any mining object does not belong to the target big data mining policy. The invoking module 330 may be configured to perform the step S130, and the detailed implementation of the invoking module 330 may refer to the detailed description of the step S130.

The updating module 340 is configured to update the target big data mining policy in the target mining source based on the target mining result to obtain a global updating result of the target big data mining policy in the target mining source, and perform big data mining based on the global updating result. The updating module 340 may be configured to perform the step S140, and the detailed implementation of the updating module 340 may refer to the detailed description of the step S140.

It should be noted that the division of the modules of the above apparatus is only a logical division, and the actual implementation may be wholly or partially integrated into one physical entity, or may be physically separated. And these modules may all be implemented in software invoked by a processing element. Or may be implemented entirely in hardware. And part of the modules can be realized in the form of calling software by the processing element, and part of the modules can be realized in the form of hardware. For example, the first obtaining module 310 may be a separate processing element, or may be integrated into a chip of the apparatus, or may be stored in a memory of the apparatus in the form of program code, and a processing element of the apparatus calls and executes the functions of the first obtaining module 310. Other modules are implemented similarly. In addition, all or part of the modules can be integrated together or can be independently realized. The processing element described herein may be an integrated circuit having signal processing capabilities. In implementation, each step of the above method or each module above may be implemented by an integrated logic circuit of hardware in a processor element or an instruction in the form of software.

Fig. 4 shows a hardware structure diagram of a digital content center 100 for implementing the cloud computing and artificial intelligence based big data mining method, as provided by the embodiment of the present disclosure, and as shown in fig. 4, the digital content center 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 (for example, the first obtaining module 310, the second obtaining module 320, the invoking module 330, and the updating module 340 included in the cloud computing and artificial intelligence based big data mining apparatus 300 shown in fig. 3), so that the processor 110 may execute the cloud computing and artificial intelligence based big data mining method according to the above method embodiment, where 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 the transceiving action of the transceiver 140, so as to perform data transceiving with the aforementioned digital content subscribing device 200.

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

In the embodiment shown in fig. 4, it should be understood that the Processor may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method disclosed in connection with the present invention may be embodied directly in a hardware processor, or in a combination of the hardware and software modules within the processor.

The machine-readable storage medium 120 may comprise high-speed RAM memory and may also include non-volatile storage NVM, such as at least one disk memory.

The bus 130 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended ISA (EISA) bus, or the like. The bus 130 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, the buses in the figures of the present application are not limited to only one bus or one type of bus.

In addition, the embodiment of the application also provides a readable storage medium, wherein the readable storage medium stores computer execution instructions, and when a processor executes the computer execution instructions, the cloud computing and artificial intelligence based big data mining method is realized.

The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

Having thus described the basic concept, it will be apparent to those skilled in the art that the foregoing detailed disclosure is to be regarded as illustrative only and not as limiting the present specification. Various modifications, improvements and adaptations to the present description may occur to those skilled in the art, although not explicitly described herein. Such modifications, improvements and adaptations are proposed in the present specification and thus fall within the spirit and scope of the exemplary embodiments of the present specification.

Also, particular push elements are used in this description to describe embodiments of this description. Reference to "one embodiment," "an embodiment," and/or "some embodiments" means that a particular feature, structure, or characteristic described in connection with at least one embodiment of the specification. Therefore, it is emphasized and should be appreciated that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, some features, structures, or characteristics of one or more embodiments of the specification may be combined as appropriate.

Moreover, those skilled in the art will appreciate that aspects of the present description may be illustrated and described in terms of several patentable species or contexts, including any new and useful combination of processes, machines, manufacture, or materials, or any new and useful improvement thereof. Accordingly, aspects of this description may be performed entirely by hardware, entirely by software (including firmware, resident software, micro-code, etc.), or by a combination of hardware and software. The above hardware or software may be referred to as "data block," module, "" engine, "" unit, "" component, "or" system. Furthermore, aspects of the present description may be represented as a computer product, including computer readable program code, embodied in one or more computer readable media.

The computer storage medium may comprise a propagated data signal with the computer program code embodied therewith, for example, on baseband or as part of a carrier wave. The propagated signal may take any of a variety of forms, including electromagnetic, optical, etc., or any suitable combination. A computer storage medium may be any computer-readable medium that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, or device. Program code located on a computer storage medium may be propagated over any suitable medium, including radio, cable, fiber optic cable, RF, or the like, or any combination of the preceding.

Computer program code required for the operation of various portions of this specification may be written in any one or more programming languages, including an object oriented programming language such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C + +, C #, VB.NET, Python, and the like, a conventional programming language such as C, VisualBasic, Fortran2003, Perl, COBOL2002, PHP, ABAP, a passive programming language such as Python, Ruby, and Groovy, or other programming languages. The program code may run entirely on the user's computer, as a stand-alone index arrangement, partly on the user's computer, partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any network format, such as a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet), or in a cloud computing environment, or as a service, such as a software as a service (SaaS).

Additionally, the order in which the elements and sequences are processed, the use of alphanumeric characters, or the use of other designations in this specification is not intended to limit the order of the processes and methods in this specification, unless otherwise specified in the claims. While various presently contemplated embodiments of the invention have been discussed in the foregoing disclosure by way of example, it is to be understood that such detail is solely for that purpose and that the appended claims are not limited to the disclosed embodiments, but, on the contrary, are intended to cover all modifications and equivalent arrangements that are within the spirit and scope of the embodiments herein. For example, although the system components described above may be implemented by hardware devices, they may also be implemented by software-only solutions, such as installing the described system on an existing server or mobile device.

Similarly, it should be noted that in the preceding description of embodiments of the present specification, various features are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure aiding in the understanding of one or more of the embodiments. This method of disclosure, however, is not intended to imply that more features than are expressly recited in a claim. Indeed, the embodiments may be characterized as having less than all of the features of a single embodiment disclosed above.

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, by way of example, and not limitation, alternative configurations of the embodiments of the specification can be considered consistent with the teachings of the specification. Accordingly, the embodiments of the present description are not limited to only those embodiments explicitly described and depicted herein.

Claims

1. A big data mining-based network training method is applied to a digital content center, wherein the digital content center is in communication connection with a plurality of digital content subscription devices, and the method comprises the following steps:

calling an initial mining network, and acquiring a prediction mining result corresponding to a first label mining source in the at least one label mining source based on original label mining source data corresponding to the first label mining source and updated reference label data corresponding to the first label mining source;

calling an initial correction network, and acquiring first correction reference degree information based on original tag mining source data corresponding to the first tag mining source and a prediction mining result corresponding to the first tag mining source;

and in response to the condition that the constraint training process does not meet the target training end condition, continuing to perform constraint training on the first correction network and the first mining network until the constraint training process meets the target training end condition, and obtaining a target correction network and a target mining network.

2. The big data mining-based network training method of claim 1, wherein the method further comprises:

acquiring a target mining source corresponding to the target big data mining strategy, original mining source data corresponding to the target mining source and updated user behavior data corresponding to the target mining source, wherein the updated user behavior data is determined based on at least part of an update result of the target big data mining strategy in the target mining source, the target mining source refers to a business item source associated with the target big data mining strategy, and the original mining source data corresponding to the target mining source refers to mining source data obtained by the target mining source when big data acquisition and last data acquisition before mining are about to be performed at this time;

3. The big data mining-based network training method according to claim 2, wherein the invoking a target mining network to obtain a target mining result corresponding to the target mining source based on the original mining source data and the updated user behavior data comprises:

calling a target mining network, and sequentially executing forward mining analysis of a first iteration strategy based on the associated feature data of the original mining source data and the updated user behavior data to obtain a first mining component corresponding to the target mining source, wherein the associated feature data refers to feature data obtained by fusing the updated user behavior data with the original mining source data, and the forward mining analysis refers to mining analysis aiming at some forward behaviors;

sequentially executing negative mining analysis of the first iteration strategy based on the target distinguishing mining features corresponding to the first mining component to obtain a second mining component corresponding to the target mining source, wherein the target distinguishing mining features corresponding to the first mining component refer to mining features obtained after target distinguishing mining is performed on all relevant mining components of the service units correspondingly covered by the first mining component, and the negative mining analysis refers to mining analysis aiming at some negative behaviors;

4. The big data mining-based network training method according to claim 2, wherein the first iteration strategy comprises a preset number of iterations of mining, the preset number of iterations is three, and any one positive mining analysis comprises one differential mining analysis and one frequent pattern mining analysis;

5. The big data mining-based network training method as claimed in claim 4, wherein any one negative mining analysis comprises one inverse discriminative mining analysis and one discriminative mining analysis; the sequentially executing negative mining analysis of the first iteration strategy based on the target distinguishing mining characteristics corresponding to the first mining component to obtain a second mining component corresponding to the target mining source comprises:

6. The big data mining-based network training method according to any one of claims 2 to 5, wherein after the target mining network is invoked and a target mining result corresponding to the target mining source is obtained based on the original mining source data and the updated user behavior data, the method further comprises:

the target correction network comprises at least one distinguishing mining network unit, at least one fusion network unit and a correction classification network unit which are sequentially connected.

7. The big data mining-based network training method according to claim 6, wherein the invoking a target correction network to obtain target correction reference degree information based on the original mining source data and the target mining result comprises:

8. The big data mining-based network training method according to any one of claims 2 to 5, wherein the target mining source is a starting at least partial mining source corresponding to the target big data mining strategy in an initial mining source; after the target big data mining strategy is updated in the target mining source based on the target mining result to obtain a global updating result of the target big data mining strategy in the target mining source, the method further includes:

9. The big data mining-based network training method according to any one of claims 2 to 5, wherein the target mining source is obtained from a service subscription mining source containing a target mining node corresponding to the target big data mining strategy;

10. A digital content center, comprising a processor, a machine-readable storage medium, and a network interface, wherein the machine-readable storage medium, the network interface, and the processor are connected via a bus system, the network interface is configured to communicatively connect with at least one digital content subscription device, the machine-readable storage medium is configured to store a program, instructions, or code, and the processor is configured to execute the program, instructions, or code in the machine-readable storage medium to perform the big data mining based network training method of any one of claims 1 to 9.