CN112597390A

CN112597390A - Block chain big data processing method based on digital finance and big data server

Info

Publication number: CN112597390A
Application number: CN202011557816.6A
Authority: CN
Inventors: 李华兵
Original assignee: Individual
Current assignee: Individual
Priority date: 2020-12-25
Filing date: 2020-12-25
Publication date: 2021-04-02

Abstract

The embodiment of the application provides a block chain big data processing method based on digital finance and a big data server, wherein a large number of user historical operation data fragments generated by non-node objects are effectively analyzed, and interest characteristic analysis is performed on target non-node objects by combining historical operation related data of corresponding target non-node objects in the user historical operation data fragments to obtain service recommendation images of the target non-node objects, so that the service recommendation can be performed on the non-node objects in a pointed manner by the node objects in a alliance block chain network. Meanwhile, the operation data of other users in the user historical operation data fragment are referred to, the service recommendation portrait of the target non-node object is obtained by integrating the use conditions of the other users on the corresponding service types, the node object can push the non-node object service more accurately, the user service experience is improved, and the pushing effect can be improved.

Description

Block chain big data processing method based on digital finance and big data server

Technical Field

The application relates to the technical field of block chains and big data analysis, in particular to a block chain big data processing method and a big data server based on digital finance.

Background

A partially decentralized federation blockchain network includes pre-designated objects, such as domain-specific authorities, e.g., large banks and financial regulators in the digital financial domain, that may become nodes in the federation blockchain network, become node objects. Other objects are not eligible to become nodes and become non-node objects. In this way, each node object in the federation blockchain network can provide a service for a non-node object, and the non-node object can interface with the federation blockchain network to request each node object in the federation blockchain network to provide a corresponding service. Existing non-node objects generate a large amount of user data during the process of using the services provided by each node object. Most of existing alliance block chain networks adopt a specific service mode to provide corresponding services for non-node objects, no effective mechanism is available for mining and analyzing massive user behavior data generated by the non-node objects, and personalized service modes aiming at different non-node objects are difficult to provide.

Disclosure of Invention

Based on the defects of the existing design, the embodiment of the application provides a block chain big data processing method based on digital finance, which is applied to a big data server, wherein the big data server is in communication connection with node objects and non-node objects in an alliance block chain network. The method comprises the following steps:

performing operation behavior tag analysis on a plurality of historical operation data fragments generated by a plurality of non-node objects by using services provided by the node objects to determine user behavior characteristic data and corresponding operation behavior tags in each historical operation data fragment;

determining a plurality of user behavior characteristic data meeting a preset data screening rule from the plurality of historical operation data segments to serve as target screening data according to the operation behavior tags of the user behavior characteristic data in each historical operation data segment;

performing target identification on historical operation data fragments generated by target non-node objects in the plurality of historical operation data fragments based on a plurality of target screening data to determine a target historical operation data fragment comprising any one target screening data;

determining a selected probability score of target screening data in each target historical operation data fragment, determining an interest degree value of the target non-node object on the target screening data according to each selected probability score, and determining target screening data which meets the interest degree value condition in the target historical operation data fragment;

and determining a service recommendation portrait of the target non-node object according to target screening data meeting a preset interest degree value condition in the target historical operation data fragments.

Further, the determining, according to the operation behavior tag of the user behavior feature data in each of the historical operation data segments, a plurality of user behavior feature data meeting a preset data filtering rule from the plurality of historical operation data segments as target filtering data includes:

determining a selected probability score of the user behavior characteristic data in each historical operation data segment according to the operation behavior tag of the user behavior characteristic data in each historical operation data segment;

and determining the user behavior characteristic data larger than a preset probability score threshold value as target screening data according to the selected probability score of the user behavior characteristic data in each historical operation data segment.

Further, the operation behavior tag of the user behavior feature data comprises a service type usage proportion, a behavior feature occurrence frequency and a service type number of people proportion; the service type usage ratio represents the ratio of the number of times of usage of the service type corresponding to the user behavior feature data in the historical operation data segment to the number of times of usage of all service types in the historical operation data segment, the behavior feature occurrence frequency represents the ratio of the number of times of occurrence of the user behavior feature data in the historical operation data segment to the number of times of occurrence of all user behavior feature data in the historical operation data segment, and the service type usage ratio represents the ratio of the number of usage of the corresponding service type in the historical user operation data segment to the number of usage of all service types in the historical user operation data segment;

determining a selected probability score of the user behavior feature data in each historical operation data segment according to the operation behavior tag of the user behavior feature data in each historical operation data segment, including:

aiming at any target user behavior characteristic data in the user behavior characteristic data in each historical operation data segment:

determining a first weight coefficient, a second weight coefficient and a third weight coefficient which respectively correspond to a service type use ratio, a behavior characteristic occurrence frequency and a service type number ratio of the target user behavior characteristic data in the historical operation data segment;

and calculating to obtain a selected probability score corresponding to the target user behavior feature data according to the corresponding service type use ratio, behavior feature occurrence frequency, service type number of people ratio, the first weight coefficient, the second weight coefficient and the third weight coefficient.

Further, the selected probability score corresponding to the target user behavior feature data is calculated by the following formula:

Score(i)＝α₁×β₁+α₂×β₂+α₃×β₃；

wherein score (i) represents the selected probability score, α, corresponding to the target user behavior feature data₁、α₂、α₃Respectively representing the service type use ratio, the behavior characteristic occurrence frequency and the service type number ratio, beta, corresponding to the behavior characteristic data of the target user₁、β₂、β₃Respectively represent the first weight coefficient, the second weight coefficient and the third weight coefficient, beta₁、β₂、β₃The sum is 1.

Further, the performing operation behavior tag analysis on the operated multiple pieces of historical operation data to determine the operation behavior tag of the user behavior feature data in each piece of historical operation data includes:

for any user behavior feature data in each historical operation data segment:

determining a service type corresponding to the user behavior characteristic data; performing first service type keyword traversal search on each historical operation data segment, and determining first occurrence times of the service types; performing second service type keyword traversal search on each historical operation data fragment, determining all service types contained in the plurality of historical operation data fragments, and determining second occurrence times of all service types; determining to obtain the service type usage ratio according to the first occurrence number and the second occurrence number;

determining user behavior characteristics corresponding to the user behavior characteristic data; performing first behavior feature keyword traversal search on each historical operation data segment, and determining third occurrence times of the user behavior features; performing second behavior feature keyword traversal search on each historical operation data segment, determining all user behavior features contained in the multiple historical operation data segments, and determining fourth occurrence times of all the user behavior features; determining to obtain the behavior feature occurrence frequency according to the third occurrence frequency and the fourth occurrence frequency;

determining a service type corresponding to the user behavior characteristic data; performing first user ID traversal search on each historical operation data fragment, and determining the number of first user IDs using the service type; performing a second service type keyword traversal search on each historical operation data fragment, determining all service types contained in the plurality of historical operation data fragments, performing a second user ID traversal search on each historical operation data fragment, and determining the number of second user IDs corresponding to all service types; and determining to obtain the service type number of people according to the number of the first user IDs and the number of the second user IDs.

Further, the determining, according to the selected probability score of the user behavior feature data in each historical operation data segment, a plurality of user behavior feature data meeting a preset data screening rule as target screening data includes:

for any of the plurality of historical operational data segments:

carrying out data denoising processing on the user behavior characteristic data in the historical operation data segment, arranging selected probability scores of the user behavior characteristic data which are left after the data denoising processing according to a descending order, and determining a first preset number of user behavior characteristic data which are arranged according to the descending order as target screening data meeting a selected probability score preset data screening rule;

the method comprises the steps of determining interest degree values of target non-node objects to target screening data according to selected probability values of the target screening data in each target historical operation data fragment, and determining service recommendation figures of the target non-node objects according to the target screening data meeting the interest degree value conditions in the target historical operation data fragments, and comprises the following steps:

for any one of the first preset number of target screening data:

weighting and summing selected probability scores of the target screening data in each target historical operation data segment to determine interest degree values of the target non-node object in the target screening data;

arranging the interest degree values of the target non-node object to each target screening data according to the sequence from large to small;

and determining the service recommended portrait of the target non-node object according to a first preset number of target screening data which are arranged from large to small.

Further, the performing operation behavior tag analysis on the operated multiple pieces of historical operation data to determine the user behavior feature data in each piece of historical operation data includes:

and respectively inputting the operated plurality of historical operation data segments into a behavior feature recognition model obtained by pre-training so as to calculate and obtain a plurality of user behavior feature data including user behavior labels in the plurality of historical operation data segments.

Further, the method further comprises: and sending the service recommendation representation of the target non-node object to all node objects in an alliance blockchain network, so that the node object deployed in the alliance blockchain network carries out node service recommendation on the target non-node object according to the service recommendation representation of the target non-node object.

The embodiment of the present application further provides a big data server, which includes a processor, and a machine-readable storage medium connected to the processor, where the machine-readable storage medium is used to store a program, an instruction, or code, and the processor is used to execute the program, the instruction, or code in the machine-readable storage medium, so as to implement the digital finance-based blockchain big data processing method according to any one of claims 1 to 9.

Compared with the prior art, the embodiment of the application has the following beneficial effects:

by effectively analyzing a large number of user historical operation data fragments generated by the non-node object and combining historical operation related data of a target non-node object corresponding to the user historical operation data fragments, interest characteristic analysis is carried out on the target non-node object to obtain a service recommendation portrait of the target non-node object, and service recommendation can be carried out on the non-node object in a pointed manner by the node object in the alliance block chain network. Meanwhile, the operation data of other users in the user historical operation data fragment is referred to, the service recommendation portrait of the target non-node object is obtained by integrating the use conditions of the other users to the corresponding service types, more accurate pushing of the service can be achieved, the service experience of the user is improved, and meanwhile the pushing effect can be improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used 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 operating environment of a federated blockchain network provided in an embodiment of the present application.

Fig. 2 is a schematic flowchart illustrating an interaction of a process of a method for processing big data of a blockchain according to an embodiment of the present disclosure.

Fig. 3 is a flowchart illustrating a sub-step of determining the service type usage percentage in step S13 shown in fig. 2.

Fig. 4 is a flow chart illustrating a sub-step of determining the occurrence frequency of the service type in step S13 shown in fig. 2.

Fig. 5 is a flowchart illustrating a sub-step of determining the service type human proportion in step S13 shown in fig. 2.

Fig. 6 is a flowchart illustrating the sub-steps of step S14 shown in fig. 2.

Fig. 7 is a flowchart illustrating the sub-steps of step S15 shown in fig. 2.

Fig. 8 is a schematic diagram of a big data server provided in an embodiment of the present application.

Detailed Description

A federated blockchain network, also known as a corporate blockchain, blockchain federated chain, or federated chain, can control the blockchain that intervenes in the consensus process through nodes pre-selected by the federated (industry) blockchain. For example, a team of 15 financial institutions, one institution operating a node, may require approval from more than 10 institutions to generate a verification block for the transaction process. The alliance blockchain network has limitations in decentralization and openness, and participants can be selected in advance. Generally, a federation blockchain network is between a public chain and a private chain. The alliance block network chain has the characteristics of openness, low trust of a public chain, privacy protection of a private block chain and single high trust.

Furthermore, a federated blockchain network differs from a public blockchain network in that the public blockchain network is completely decentralized and transparent to outside disclosure, and any subject or object (e.g., person or entity) can become a node or an object in the public blockchain network. The alliance block chain network is partially decentralized and not exposed to the outside, and only a pre-designated subject or object (an authority in a certain field, such as a large bank and a financial supervision institution in the financial field) can become a node in the alliance block chain network and become a node object. Other principals are not eligible to become nodes and become non-node objects that the federated blockchain network can serve, which can interface with the federated blockchain network requesting the federated blockchain network to achieve a legal consensus for the generated traffic.

Based on the problems described in the foregoing background art, most of the existing alliance blockchain networks only provide corresponding services for non-node objects based on existing service modes, and there is no effective mechanism to analyze and filter big data generated by the non-node objects, which is difficult to provide personalized services for personalized users. In other big data fields, there are some big data analysis methods for user portraits, which can perform portraits analysis on historical data of users, thereby providing interesting service push for users. However, most of these portrait analyzing methods perform a portrait of interest based on operation data of a single user. In a specific service scene of an alliance block chain network, most authoritative node main bodies such as financial institutions, securities and insurance banks provide services for non-node main bodies, if big data analysis is performed only on a certain specific user (or a specific non-node object), the popularity of main service businesses of the node main bodies and the popularity of each service business in a large user group cannot be considered, and further an analysis result formed by the big data analysis on a single user is caused, so that the popularization of the main service businesses is not facilitated in subsequent application, and the pushing accuracy of the service businesses is difficult to guarantee.

Based on the discovery of the foregoing problems, the inventor of the present application innovatively provides the following technical solutions to implement big data analysis and processing of a blockchain network, refers to operation data of other users in a user historical operation data fragment, and synthesizes usage situations of the corresponding service types by the other users to obtain a service recommendation sketch of a target non-node object, so as to consider historical data of the target non-node object, and also refer to related historical data of the other users for different service types of services to obtain a service recommendation sketch of the target non-node object, so that not only are data such as usage interests and habits of the target node object taken into account, but also factors such as probability of different service types of services of the node object being selected by different users are combined, and more accurate pushing of services can be implemented, and service experience of the users is improved.

In detail, an embodiment of the present application provides a method for processing big data of a block chain, which is applied to a big data server, where the big data server is in communication connection with a node object and a non-node object in a federated block chain network. The method comprises the following steps:

Specific implementation manners of the steps of the above method will be described in detail below with reference to the drawings and the embodiments.

Referring to fig. 1, fig. 1 is a schematic diagram of an operating environment according to an embodiment of the present application, where the operating environment relates to a federated blockchain network 10, and the federated blockchain network 10 includes a big data server 100 and a plurality of non-node objects 200 and a plurality of node objects 100 communicatively connected to the big data server 100. Node objects within the federated blockchain network 10 may serve non-node objects 300. The big data server 100 may be a big data analysis server where the node objects 300 in the federation blockchain network 10 achieve security consensus establishment, or in other embodiments, one of the node objects 300 may play the role of the big data server 100, which is not limited herein. In the embodiment of the present application, the big data server 100 may perform big data processing, analysis, data mining and other operations on a large number of historical user operation data fragments generated by a large number of non-node objects 200, so as to provide the node objects 300 with targeted personalized services or recommendations and the like for different non-node objects 200. The detailed method is specifically exemplified below.

Referring to fig. 2, a schematic flow interaction diagram of a block chain big data processing method according to an embodiment of the present application is shown, where the block chain big data processing method is mainly executed by the big data server 100 in the block chain network 10, and is specifically described as follows.

In step S01, the history operation data segment is transmitted. Specifically, the non-node object 200 transmits a large number of pieces of historical operation data, which are generated when using the service provided by the node object 300, to the big data server 100 to be analyzed and processed.

Step S11, the behavior tag analysis is operated.

Specifically, the big data server 100 performs an operation behavior tag analysis on a plurality of historical operation data segments sent by a plurality of non-node objects 200 and generated when the services provided by the node objects 300 are used, so as to determine user behavior feature data and corresponding operation behavior tags in each historical operation data segment.

In this embodiment, the specific manner of performing operation behavior tag analysis on a plurality of operated historical operation data segments to determine the user behavior feature data in each of the historical operation data segments may be: and respectively inputting the operated plurality of historical operation data segments into a behavior feature recognition model obtained by pre-training so as to calculate and obtain a plurality of user behavior feature data including user behavior labels in the plurality of historical operation data segments.

In this embodiment, the historical operation data segment may refer to data generated in a preset time period, for example, in the last week or the last month, and is stored in the big data server 100 in the form of a data segment, or may be one or more historical operation data segments formed by data generated by a user of the non-node object 200 during each business operation. The user behavior feature data may be feature data for characterizing a user's specific operation, such as purchasing a financial product, consulting insurance services, transferring money, making loans, repayment, making financing requests, etc., when the user uses the service provided by the node object 300. The user behavior feature data may be, for example, a business behavior category including an operation attribute of true and corresponding business data information, such as a data field for indicating that "user a successfully purchases fund product C". The operation behavior tag may be used to represent specific operation information of a user in the historical operation data segment, and may be, for example, a data tag including "operation mode", "operation frequency", "generation amount" and the like for a certain service type, and the data tag may exist in any machine-recognizable form, such as a data index tag, a data table, a data field, and the like.

And step S12, screening the user behavior characteristic data.

In detail, the big data server 100 determines, from the plurality of historical operation data segments, a plurality of user behavior feature data satisfying a preset data filtering rule as target filtering data according to the operation behavior tag of the user behavior feature data in each of the historical operation data segments. In this embodiment, a plurality of user behavior feature data meeting the preset data screening rule are determined from a large number of historical operation data segments and used as target screening data, which is beneficial to subsequent targeted identification of data (effective feature data) with obvious features, and can improve the data processing efficiency.

Specifically, in step S12, according to the operation behavior tag of the user behavior feature data in each of the historical operation data segments, a plurality of user behavior feature data satisfying a preset data filtering rule are determined from the plurality of historical operation data segments as target filtering data, and a specific implementation manner may be:

firstly, determining a selected probability score of user behavior characteristic data in each historical operation data segment according to an operation behavior label of the user behavior characteristic data in each historical operation data segment;

and then, according to the selected probability score of the user behavior characteristic data in each historical operation data segment, determining the user behavior characteristic data larger than a preset probability score threshold value as target screening data.

In this way, the selected probability score can be determined according to the category of the operation behavior tag of the user behavior feature data and the big data analysis result, for example, in the operation data of ten thousand people, if the ratio of the number of times that a certain operation behavior tag is operated to the total number of times is 35%, the selected probability score can be 35. In the final screening process, the user behavior feature data with the selected probability score exceeding 60 points can be selected as the target screening data. Therefore, the operation characteristics of a specific user and the operation characteristics of other users in the big data to a certain service are considered in the target screening data, the pushing effect can be better for the subsequent service pushing executed again according to data analysis, and the probability of selective use of other users of the pushed service is higher.

In step S13, target recognition is performed on the historical operation data segment.

In detail, in this embodiment of the application, the big data server 100 performs target identification on a historical operation data segment generated by a target non-node object in the plurality of historical operation data segments based on a plurality of target screening data, so as to determine a target historical operation data segment including any one of the target screening data. The specific method may be that the target screening data is sequentially and respectively subjected to feature recognition or data comparison with each historical operation data segment, and when any one of the historical operation data segments includes corresponding target screening data, the historical operation data segment is determined as one target historical operation data segment. And if any one target screening data is not included in one historical operation data segment, the historical operation data segment is not a target historical operation data segment.

In step S14, target screening data satisfying the interest level condition is determined.

In detail, in this embodiment of the application, after the target screening data is determined, the big data server 100 further determines a selected probability score of the target screening data in each target historical operation data segment, determines an interest degree value of the target non-node object in the target screening data according to each selected probability score, and then determines the target screening data in the target historical operation data segment that meets the interest degree value condition. The specific method may be to compare the interest degree value with a preset interest degree value interval, and if the interest degree value is within the interest degree value interval, it may be determined that the target screening data having the interest degree value satisfies the interest degree value condition. Or, the interest degree values of all the target screening data may be sorted, and the target screening data sorted in the preset number of bits before may be determined to satisfy the interest degree value condition.

Step S15, determining the service recommendation portrait of the target non-node object.

In detail, in this embodiment of the application, the big data server 100 determines the service recommendation representation of the target non-node object according to the target screening data meeting the preset interest level condition in the target historical operation data segment. For example, the historical operation data segment generated by the target non-node object may be compared with all the target screening data meeting the interest level value condition, any one or more target screening data meeting the interest level value condition included in the historical operation data segment may be found, and the service recommendation representation of the target non-node object may be determined according to the found any one or more target screening data. For example, if the searched target screening data includes data related to "purchase fund product C", "consult loan service", "purchase serious insurance", etc., the service recommendation for identifying the target non-node object may be represented by: "fund product, loan service, critical insurance", but not limited thereto.

Step S16, the service recommends portrait pushing.

In detail, in this embodiment, after determining the service recommendation representation of any target non-node object, the big data server 100 may push the service recommendation representation of the target non-node object to the node object 300, so that the node object 300 may recommend a node service for the target non-node object according to the service recommendation representation of the target non-node object.

And step S21, node service recommendation is carried out for the target non-node object.

In detail, in this embodiment, after receiving the service recommendation representation of any target non-node object, the node object 300 may recommend a node service to the target non-node object when the target non-node object initiates a service connection to the node object 300. For example, the operation item of the corresponding service corresponding to the service recommendation image may be preferentially displayed at a set significant position on a service interface provided for the target non-node object, but not limited thereto.

Further, in this embodiment, the operation behavior tag of the user behavior feature data may include a service type usage percentage, a behavior feature occurrence frequency, and a service type number of people percentage. The service type usage percentage represents the ratio of the number of times of usage of the service type corresponding to the user behavior feature data in the historical operation data segment to the number of times of usage of all service types in the historical operation data segment, the behavior feature occurrence frequency represents the ratio of the number of times of occurrence of the user behavior feature data in the historical operation data segment to the number of times of occurrence of all user behavior feature data in the historical operation data segment, and the service type usage percentage represents the ratio of the number of usage of the corresponding service type in the historical user operation data segment to the number of usage of all service types in the historical user operation data segment.

Further, in step S13, performing operation behavior label analysis on the operated multiple pieces of historical operation data to determine operation behavior labels of user behavior feature data in each piece of historical operation data, including a step of determining to obtain the service type usage percentage, a step of determining the occurrence frequency of the behavior feature, and a step of determining the service type human occupation ratio, respectively. In detail, referring to fig. 3, the step of determining the service type usage percentage specifically includes the following sub-steps.

And a substep S1301, aiming at any user behavior characteristic data in each historical operation data fragment, determining a service type corresponding to the user behavior characteristic data.

And a substep S1302, performing a first service type keyword traversal search on each historical operation data segment, and determining a first occurrence number of the service type.

And a substep S1303, performing a second service type keyword traversal search on each historical operation data segment, determining all service types contained in the plurality of historical operation data segments, and determining second occurrence times of all service types.

And a substep S1304, determining to obtain the service type usage ratio according to the first occurrence number and the second occurrence number.

Further, referring to fig. 4, the step of determining the occurrence frequency of the service type specifically includes the following sub-steps.

And a substep S1311, determining a user behavior characteristic corresponding to the user behavior characteristic data.

And a substep S1312 of performing a first behavior feature keyword traversal search on each historical operation data segment, and determining a third occurrence number of the user behavior feature.

Step S1313, performing second behavior feature keyword traversal search on each historical operation data segment, determining all user behavior features included in the multiple historical operation data segments, and determining fourth occurrence times of all user behavior features.

Step S1314, determining to obtain the behavior feature occurrence frequency according to the third occurrence frequency and the fourth occurrence frequency. The behavior feature frequency of occurrence may be a ratio of the third number of occurrences to the fourth number of occurrences.

Further, referring to fig. 5, the step of determining the service type human proportion specifically includes the following sub-steps.

And a substep S1321, determining a service type corresponding to the user behavior feature data.

And a substep S1322 of performing a first user ID traversal search on each of the historical operation data fragments, and determining a first user ID number using the service type.

And a substep S1323, performing a second service type keyword traversal search on each historical operation data segment, determining all service types included in the plurality of historical operation data segments, performing a second user ID traversal search on each historical operation data segment, and determining the number of second user IDs corresponding to all service types. For example, traversal search may be performed in the historical operation data segment according to all preset service type keywords to find out all service types, then the historical operation data segment is traversed sequentially for each service type to obtain the number of user IDs corresponding to each service type, and then the number of user IDs is summed to obtain the second user ID number.

And a substep S1324 of determining to obtain the service type human number ratio according to the first user ID number and the second user ID number.

Further, in step S14, determining the selected probability score of the user behavior feature data in each historical operation data segment according to the operation behavior label of the user behavior feature data in each historical operation data segment may be implemented in the following manner.

Firstly, aiming at any target user behavior characteristic data in the user behavior characteristic data in each historical operation data segment, determining a first weight coefficient, a second weight coefficient and a third weight coefficient which respectively correspond to a service type use ratio, behavior characteristic occurrence frequency and service type human frequency ratio of the target user behavior characteristic data in the historical operation data segment;

and then, calculating to obtain a selected probability score corresponding to the target user behavior feature data according to the corresponding service type use ratio, behavior feature occurrence frequency, service type number of people ratio, the first weight coefficient, the second weight coefficient and the third weight coefficient.

Illustratively, the selected probability score corresponding to the target user behavior feature data is calculated by the following formula:

Score(i)＝α₁×β₁+α₂×β₂+α₃×β₃；

wherein score (i) represents the selected probability score, α, corresponding to the target user behavior feature data₁、α₂、α₃Respectively representing the service type use ratio, the behavior characteristic occurrence frequency and the service type number ratio, beta, corresponding to the behavior characteristic data of the target user₁、β₂、β₃Respectively represent the first weight coefficient, the second weight coefficient and the third weight coefficient,β₁、β₂、β₃the sum is 1.

Further, in the step S14, a manner of determining, according to the selected probability score of the user behavior feature data in each of the historical operation data segments, a plurality of user behavior feature data that satisfy the preset data filtering rule as target filtering data may be specifically implemented in a manner shown in fig. 6, which is specifically described below.

And a substep S141, performing data denoising processing on the user behavior feature data in the historical operation data segment for any historical operation data segment in the plurality of historical operation data segments.

And a substep S142, arranging the selected probability scores of the residual user behavior characteristic data after the data denoising treatment according to a descending order.

And a substep S143, determining the first preset number of user behavior characteristic data after being arranged from big to small as target screening data meeting the selected probability score preset data screening rule.

Further, in step S15, an interest degree value of the target non-node object for the target screening data is determined according to the selected probability score of the target screening data in each target historical operation data segment, and a service recommendation image of the target non-node object is determined according to the target screening data satisfying the interest degree value condition in the target historical operation data segment, which may be implemented by referring to the method shown in fig. 7, and is described in detail as follows.

And a substep S151, for any target screening data in the first preset number of target screening data, performing weighted summation on the selected probability scores of the target screening data in each target historical operation data segment to determine interest degree values of the target non-node targets in the target screening data.

And a substep S152, arranging the interest degree values of the target non-node object to each target screening data according to a descending order.

And a substep S153, determining the service recommendation portrait of the target non-node object according to the first preset number of target screening data which are arranged from big to small. For example, the service types corresponding to the first preset number of target screening data may be used as the service recommendation image, and when the non-node object 200 uses the service provided by the node object 300, the operation items of the service types corresponding to the service recommendation image may be preferentially displayed on the main page of the service.

Further, in this embodiment of the application, after determining the service recommendation representation of the target non-node object, the big data server 100 itself may serve as a node object to push the matched service recommendation information for the target non-node object. The recommended method may be as follows:

acquiring target data source configuration information to be configured regularly according to the service recommended representation of the target non-node object, and performing link object index on the target data source configuration information to obtain target link object index representation, wherein the target link object index representation comprises target link object distribution corresponding to the target data source configuration information;

acquiring target subject distribution in the target data source configuration information, and determining first expansion subject distribution corresponding to the target subject distribution;

determining target interest hotspot word embedding characteristics corresponding to the target topic distribution according to the topic distribution embedding characteristics of the first extended topic distribution and corresponding topic belonging relation parameters, wherein the topic belonging relation parameters are obtained according to the association degree between the target link object distribution and the extended relation embedding characteristics, and the extended relation embedding characteristics are embedding characteristic vectors representing topic distribution extended relations;

clustering target interest hot word embedding characteristics corresponding to the target topic distribution represented by the target link object index to obtain target clustering characteristics;

and determining a data source configuration information processing result corresponding to the target data source configuration information according to the target clustering characteristics.

In summary, the data push model may include a link object index model of the data source configuration information, an interest hot word vector extraction network, and a clustering model, where the interest hot word vector extraction network may determine, according to the topic distribution embedding characteristics and the corresponding topic affiliation parameters, tagged interest hot word embedding characteristics corresponding to the training topic distribution, and the topic affiliation parameters are obtained according to the association between the training link object distribution and the expansion relationship embedding characteristics, where the expansion relationship embedding characteristics are embedding characteristic vectors representing the topic distribution expansion relationship. In addition, the topic affiliation parameter is obtained according to the degree of association between the training link object distribution and the expansion relationship embedded feature, so that the importance degree of the topic distribution embedded feature of the expansion topic distribution to the embedded feature of the training topic distribution can be determined according to the topic distribution feature of the sample data source configuration information, and the topic affiliation parameter is determined according to the importance degree, so that the obtained annotation interest hot word vector can better promote the understanding of the topic distribution feature of the sample data source configuration information, the data source configuration information understanding ability of the obtained data push model and the preset network model is improved, and the accuracy of the data source configuration information processing result is improved.

Further, the determining of the target interest hotspot word embedding feature corresponding to the target topic distribution according to the topic distribution embedding feature of the first expanded topic distribution and the corresponding topic affiliation parameter may specifically be implemented by the following steps:

obtaining a relation theme label model formed by the first expansion theme distribution and the target theme distribution;

for the theme distribution of the theme label model in the relation theme label model, acquiring an expansion relation embedding characteristic which represents the expansion relation between the theme distribution of the theme label model and the inheritance theme distribution;

obtaining topic contact degree according to the expansion relation embedding characteristics and the target link object distribution, and determining a topic belonging relation parameter corresponding to the inheritance topic distribution according to the topic contact degree;

determining target interest hotspot word embedding characteristics corresponding to the theme distribution of the theme label model according to the theme belonging relation parameters corresponding to the inherited theme distribution and the theme distribution embedding characteristics of the inherited theme distribution;

extracting target interest hotspot word embedding characteristics corresponding to the target topic distribution from the target interest hotspot word embedding characteristics corresponding to the topic distribution of each topic label model of the relational topic label model.

In addition, the target interest hot word embedding characteristics corresponding to the topic distribution of the topic tag model are output by an interest hot word vector extraction network, the interest hot word vector extraction network includes at least one target embedding extraction unit, and the target interest hot word embedding characteristics corresponding to the topic distribution of the topic tag model are determined according to the topic belonging relation parameter corresponding to the inheritance topic distribution and the topic distribution embedding characteristics of the inheritance topic distribution, which can be realized by the following steps:

inputting the theme distribution embedding characteristics inheriting the theme distribution and the expansion relation embedding characteristics into the target embedded extraction unit for processing to obtain first interest hot word embedding characteristics corresponding to the theme distribution of the theme label model;

and determining target interest hotspot word embedding characteristics corresponding to the theme distribution of the theme label model according to the first interest hotspot word embedding characteristics corresponding to the theme distribution of the theme label model and the corresponding theme belonging relationship parameters corresponding to the inherited theme distribution.

Meanwhile, the theme distribution embedding feature inheriting the theme distribution and the expansion relation embedding feature are input into the target embedded extraction unit to be processed, so that a first interest hotspot word embedding feature corresponding to the theme distribution of the theme label model is obtained, and the method can be realized in the following manner:

determining a target theme expansion trend according to a theme distribution expansion relation between the theme distribution and the inheritance theme distribution of the theme label model, wherein the target theme expansion trend is expansion enhancement or expansion weakening;

calculating the theme distribution embedding characteristics of the inherited theme distribution and the expansion relation embedding characteristics according to the target theme expansion trend to obtain the calculation embedding characteristics corresponding to the theme distribution of the theme label model;

and processing the calculated embedding characteristics by using the parameters of the embedded extraction unit in the target embedded extraction unit to obtain the first interest hot word embedding characteristics corresponding to the theme distribution of the theme label model.

In addition, the processing of the calculated embedding features by using the parameters of the embedded extraction unit in the target embedded extraction unit to obtain the first interest hotspot word embedding features corresponding to the theme distribution of the theme tag model can be realized by the following steps:

acquiring output embedded characteristics corresponding to the distribution of the inherited topics, which are output by a last embedded extraction unit corresponding to the target embedded extraction unit in the interest hot word vector extraction network; and the target embedded extraction unit processes the calculation embedding characteristics and the output embedding characteristics by using parameters of a first embedded extraction unit to obtain first interest hot word embedding characteristics corresponding to the theme distribution of the theme label model.

In this embodiment, the target interest hot word embedding characteristics corresponding to the topic distribution of the topic tag model are output by an interest hot word vector extraction network, and the interest hot word vector extraction network includes at least one target embedding extraction unit. Based on this, the topic contact degree is obtained according to the distribution of the expansion relationship embedded feature and the target link object, and the topic belonging relationship parameter corresponding to the inheritance topic distribution is determined according to the topic contact degree, which can be realized by the following method:

processing the extended relation embedded feature by using a second embedded extraction unit parameter in the target embedded extraction unit to obtain a subject dimension feature;

processing the target link object distribution by using a third embedded extraction unit parameter in the target embedded extraction unit to obtain a theme quantization characteristic;

calculating according to the theme dimension characteristics and the theme quantization characteristics to obtain theme contact degree;

and determining subject affiliation parameters corresponding to the inheritance subject distribution according to the subject contact degree, wherein the subject contact degree and the subject affiliation parameters corresponding to the inheritance subject distribution form a positive correlation.

In addition, in this embodiment, the target data source configuration information includes a plurality of configuration service, the target link object index representation includes a link object index representation sequence of the configuration service, and the link object index representation sequence of the configuration service includes a link object index representation sequence of a configuration service corresponding to each configuration service. Based on this, the target linked object index represents the target interest hotspot word embedding characteristics corresponding to the target theme distribution, and the target clustering characteristics are obtained through clustering, which can be realized through the following steps:

performing interest hot word mapping processing on a link object index representation sequence of the configuration service corresponding to the target configuration service according to the target interest hot word embedding characteristics corresponding to the target theme distribution to obtain a link object index representation of the interest hot word mapping corresponding to the target configuration service;

updating the link object index representation sequence of the configuration service in the link object index representation sequence of the configuration service by using the link object index representation mapped by the interest hotspot word corresponding to the target configuration service, and obtaining the updated link object index representation sequence of the configuration service;

and clustering the updated link object index representation sequence of the configuration service and the target link object distribution by using a clustering model to obtain the link object index representation sequence of the configuration service after clustering and the target link object distribution after clustering.

Further, in this embodiment, the determining, according to the target clustering characteristic, a data source configuration information processing result corresponding to the target data source configuration information may specifically be to input the clustered target link objects into a trained update model in a distributed manner, so as to obtain a data source configuration information update result corresponding to the target data source configuration information.

In this embodiment, the performing, according to the target interest hotspot word embedding feature corresponding to the target theme distribution, interest hotspot word mapping processing on the link object index representation sequence of the configuration service corresponding to the target configuration service to obtain a link object index representation of interest hotspot word mapping corresponding to the target configuration service may specifically be: and fusing the target interest hotspot word embedding characteristics corresponding to the target theme distribution with the link object index representation sequence of the configuration service corresponding to the target configuration service to obtain the link object index representation mapped by the interest hotspot words corresponding to the target configuration service.

On the basis of the above content, in the embodiment of the application, when the service portrait recommendation is performed, the network parameters of the used related data push model can be adjusted through the following steps to further improve the recommendation accuracy. The corresponding adjustment method is described in detail as follows:

acquiring sample data source configuration information and a standard data source configuration information processing result corresponding to the sample data source configuration information;

inputting the sample data source configuration information into a link object index model of the data source configuration information to obtain a sample data source configuration information link object index representation, wherein the sample data source configuration information link object index representation comprises training link object distribution corresponding to the sample data source configuration information;

acquiring training subject distribution corresponding to the sample data source configuration information, and determining second expansion subject distribution corresponding to the training subject distribution;

inputting the topic distribution embedding features corresponding to the second expansion topic distribution into an interest hot word vector extraction network, and determining the marked interest hot word embedding features corresponding to the training topic distribution according to the topic distribution embedding features and corresponding topic affiliation parameters, wherein the topic affiliation parameters are obtained according to the association degree between the training link object distribution and the expansion relation embedding features, and the expansion relation embedding features are embedding feature vectors representing topic distribution expansion relations;

inputting the embedding characteristics of the marked interest hot words, which are expressed by the sample data source configuration information link object indexes and correspond to the training topic distribution, into a clustering model for clustering to obtain a training clustering result;

processing the training clustering result according to a preset network model to obtain a training processing result;

adjusting parameters of the preset network model or adjusting parameters of the preset network model and a data pushing model according to the training processing result and the standard data source configuration information processing result, wherein the data pushing model comprises a link object index model of the data source configuration information, the interest hot word vector extraction network and the clustering model.

Based on the same inventive concept, please refer to fig. 8, which shows a schematic block diagram of a big data server 100 for executing the above block chain big data processing method according to an embodiment of the present application, where the big data server 100 may include a block chain big data processing system 110, a machine-readable storage medium 120, and a processor 130.

In this embodiment, the machine-readable storage medium 120 and the processor 130 may be located in the big data server 100 and separately located. The machine-readable storage medium 120 may also be separate from the big data server 100 and may be accessed by the processor 130 through a bus interface. Alternatively, the machine-readable storage medium 120 may be integrated into the processor 130, e.g., may be a cache and/or general purpose registers.

The blockchain big data processing system 110 may include software functional modules stored in the machine readable storage medium 120 (e.g., the blockchain big data processing system 110 includes the respective software functional modules. when the processor 130 executes the software functional modules in the blockchain big data processing system 110, the blockchain big data processing method provided by the foregoing method embodiments is implemented.

In detail, the blockchain big data processing system 110 includes a plurality of software functional modules, such as a behavior tag analysis module 111, a feature data filtering module 112, a target data identification module 113, an interest data determination module 114, and a service recommendation representation module 115.

The behavior tag analysis module 111 is configured to perform operation behavior tag analysis on a plurality of historical operation data segments generated by a plurality of non-node objects using services provided by node objects, so as to determine user behavior feature data and corresponding operation behavior tags in each of the historical operation data segments. It is to be understood that the behavior tag analysis module 111 can be configured to perform the step S11, and for the detailed implementation of the behavior tag analysis module 111, reference may be made to the content related to the step S11, and details are not repeated here.

The characteristic data screening module 112 is configured to determine, according to the operation behavior tag of the user behavior characteristic data in each historical operation data segment, a plurality of user behavior characteristic data meeting a preset data screening rule from the plurality of historical operation data segments to serve as target screening data. It is understood that the feature data filtering module 112 may be configured to perform the step S12, and the detailed implementation of the feature data filtering module 112 may refer to the content related to the step S12, which is not repeated herein.

The target data identification module 113 is configured to perform target identification on a historical operation data segment generated by a target non-node object in the plurality of historical operation data segments based on a plurality of target screening data, so as to determine a target historical operation data segment including any one of the target screening data. It is understood that the target data identification module 113 may be configured to perform the step S13, and details regarding the implementation of the target data filtering module 113 may refer to the above-mentioned contents related to the step S13, which is not repeated herein.

The interest data determining module 114 is configured to determine a selected probability score of the target screening data in each target historical operation data segment, determine an interest degree value of the target non-node object in the target screening data according to each selected probability score, and determine the target screening data in the target historical operation data segment that meets the interest degree value condition. It is understood that the interest data determining module 114 may be configured to perform the step S14, and the detailed implementation of the interest data determining module 114 may refer to the content related to the step S14, which is not repeated herein.

The service recommendation portrait module 115 is configured to determine a service recommendation portrait of the target non-node object according to target screening data that satisfies a preset interest level value condition in the target historical operation data segment. It is understood that the service recommendation image module 115 can be used to execute the step S15, and the detailed implementation of the service recommendation image module 115 can refer to the content related to the step S15, which is not repeated herein.

In summary, the big data processing method and the big data server for the block chain provided in the embodiment of the present application perform effective analysis on a large number of user historical operation data fragments generated by a non-node object, and perform interest characteristic analysis on a target non-node object by combining historical operation related data of a corresponding target non-node object in the user historical operation data fragments to obtain a service recommendation representation of the target non-node object, so that the service recommendation of the non-node object with respect to the node object in the alliance block chain network can be facilitated. Meanwhile, the operation data of other users in the user historical operation data fragment are referred to, the service recommendation portrait of the target non-node object is obtained by integrating the use conditions of the other users on the corresponding service types, the node object can push the non-node object service more accurately, the service experience of the user is improved, and the pushing effect can be improved.

The embodiments described above are only a part of the embodiments of the present application, and not all of the embodiments. The components of the embodiments of the present application, as generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations. Thus, the detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the application, but is merely representative of selected embodiments of the application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims. Moreover, all other embodiments that can be made available by a person skilled in the art without making any inventive step based on the embodiments of the present application shall fall within the scope of protection of the present application.

Claims

1. A blockchain big data processing method based on digital finance is applied to a big data server which is in communication connection with node objects and non-node objects in a alliance blockchain network, and the method comprises the following steps:

2. The method according to claim 1, wherein the determining, from the plurality of historical operation data segments, a plurality of user behavior feature data meeting a preset data filtering rule as target filtering data according to the operation behavior tag of the user behavior feature data in each of the historical operation data segments comprises:

3. The method of claim 2, wherein the operation behavior tags of the user behavior feature data comprise service type usage percentage, behavior feature occurrence frequency, service type popularity percentage; the service type usage ratio represents the ratio of the number of times of usage of the service type corresponding to the user behavior feature data in the historical operation data segment to the number of times of usage of all service types in the historical operation data segment, the behavior feature occurrence frequency represents the ratio of the number of times of occurrence of the user behavior feature data in the historical operation data segment to the number of times of occurrence of all user behavior feature data in the historical operation data segment, and the service type usage ratio represents the ratio of the number of usage of the corresponding service type in the historical user operation data segment to the number of usage of all service types in the historical user operation data segment;

4. The method of claim 2, wherein the selected probability score corresponding to the target user behavior feature data is calculated by the following formula:

Score(i)＝α₁×β₁+α₂×β₂+α₃×β₃；

5. The method of claim 3, wherein performing operation behavior tag analysis on the operated plurality of historical operation data segments to determine operation behavior tags of the user behavior feature data in each of the historical operation data segments comprises:

for any user behavior feature data in each historical operation data segment:

6. The method according to claim 2, wherein determining a plurality of user behavior feature data satisfying a preset data filtering rule as target filtering data according to the selected probability scores of the user behavior feature data in each of the historical operation data segments comprises:

for any of the plurality of historical operational data segments:

for any one of the first preset number of target screening data:

7. The method as claimed in claim 1, wherein the performing operation behavior tag analysis on the operated plurality of historical operation data segments to determine user behavior feature data in each of the historical operation data segments comprises:

8. The method according to any one of claims 1 to 7, further comprising:

and sending the service recommendation representation of the target non-node object to all node objects in an alliance blockchain network, so that the node object deployed in the alliance blockchain network carries out node service recommendation on the target non-node object according to the service recommendation representation of the target non-node object.

9. The method according to any one of claims 1 to 8, further comprising:

pushing matched service recommendation information to the target non-node object according to the service recommendation image of the target non-node object;

wherein the step of pushing the matched service recommendation information to the target non-node object according to the service recommendation picture of the target non-node object comprises:

10. A big data server, comprising a processor, a machine-readable storage medium connected with the processor, the machine-readable storage medium storing a program, instructions or code, and the processor executing the program, instructions or code in the machine-readable storage medium to implement the method for processing big data based on digital finance block chain according to any one of claims 1-9.