CN110033092B

CN110033092B - Data label generation method, data label training device, event recognition method and event recognition device

Info

Publication number: CN110033092B
Application number: CN201910095815.5A
Authority: CN
Inventors: 程羽; 刘腾飞; 王维强; 杨洋; 徐轶
Original assignee: Alibaba Group Holding Ltd
Current assignee: Advanced New Technologies Co Ltd; Advantageous New Technologies Co Ltd
Priority date: 2019-01-31
Filing date: 2019-01-31
Publication date: 2020-06-02
Anticipated expiration: 2039-01-31
Also published as: TWI713019B; TW202038217A; WO2020155831A1; CN110033092A

Abstract

The embodiment of the application provides a method and a device for generating a data label, training a model and identifying an event, wherein the method for generating the data label comprises the following steps: acquiring a plurality of pieces of event data of a target event; respectively processing each piece of event data through a plurality of event identification strategies corresponding to the target event to obtain first identification result data of each piece of event data relative to the risk event; performing data inference based on the first recognition result data according to a data inference model corresponding to the first recognition result data to obtain second recognition result data of each event data relative to the risk event; and determining a data label of each piece of event data according to the second identification result data of each piece of event data.

Description

Data label generation method, data label training device, event recognition method and event recognition device

Technical Field

The application relates to the field of computer equipment, in particular to a method and a device for generating a data label, training a model and identifying an event.

Background

With the development of artificial intelligence technology and deep learning technology, various models can be built in a computer, and various events are processed through the models. For example, the user complaint event is processed through the user complaint event recognition model to judge whether the user complaint event is established. The customer complaint event can be a complaint of other customers' gambling, money laundering, etc.

The main process of training the model at present is as follows: and obtaining sample data, labeling the sample data, and training the model by using the labeled sample data. At present, data are mainly labeled in a manual mode, and the problem that the accuracy of the label is difficult to guarantee exists, so that the accuracy of a data processing result of a model is low.

Disclosure of Invention

The embodiment of the application aims to provide a data label generation method, a model training method and an event recognition device, so as to solve the problem that label accuracy is difficult to guarantee in manual label beating, and improve accuracy of a data processing result of a model.

In order to solve the above technical problem, the embodiment of the present application is implemented as follows:

the embodiment of the application provides a data tag generation method, which comprises the following steps:

acquiring a plurality of pieces of event data of a target event;

respectively processing each piece of event data through a plurality of event identification strategies corresponding to the target event to obtain first identification result data of each piece of event data relative to the risk event;

performing data inference based on the first recognition result data according to a data inference model corresponding to the first recognition result data to obtain second recognition result data of each event data relative to the risk event;

and determining a data label of each piece of event data according to the second identification result data of each piece of event data.

The embodiment of the application provides a model training method, which comprises the following steps:

determining a data label of event data of a target event according to the data label generating method;

taking the event data of the target event as sample data of the target event;

training a risk event identification model corresponding to the target event according to the sample data of the target event and the data label of the sample data, wherein the risk event identification model is used for identifying whether the target event is a risk event or not, or identifying the probability that the target event is a risk event.

The embodiment of the application provides an event identification method, which comprises the following steps:

acquiring event data of a target event to be identified;

processing the event data of the target event to be recognized by utilizing a risk event recognition model obtained by training through the model training method;

and determining whether the target event to be identified is a risk event or not according to the processing result, or determining the probability that the target event to be identified is the risk event.

An embodiment of the present application provides a data tag generation apparatus, including:

the first acquisition module is used for acquiring a plurality of pieces of event data of a target event;

the first result determining module is used for respectively processing each piece of event data through a plurality of event identification strategies corresponding to the target event to obtain first identification result data of each piece of event data relative to the risk event;

a second result determining module, configured to perform data inference based on the first recognition result data according to a data inference model corresponding to the first recognition result data, to obtain second recognition result data of each piece of event data relative to a risk event;

and the label generation module is used for determining the data label of each piece of event data according to the second identification result data of each piece of event data.

The embodiment of the application provides a model training device, includes:

a tag determination module, configured to determine a data tag of event data of a target event according to the data tag generation apparatus;

a sample determining module, configured to use event data of the target event as sample data of the target event;

and the model training module is used for training a risk event identification model corresponding to the target event according to the sample data of the target event and the data label of the sample data, wherein the risk event identification model is used for identifying whether the target event is a risk event or not, or identifying the probability that the target event is a risk event.

An embodiment of the present application provides an event recognition apparatus, including:

the second acquisition module is used for acquiring event data of the target event to be identified;

the data processing module is used for processing the event data of the target event to be recognized by utilizing the risk event recognition model obtained by training of the model training device;

and the event identification module is used for determining whether the target event to be identified is a risk event or not according to the processing result, or determining the probability that the target event to be identified is the risk event.

An embodiment of the present application provides an electronic device, including: a processor; and a memory arranged to store computer executable instructions that, when executed, cause the processor to implement the steps of the data tag generation method described above, or the steps of the model training method described above, or the steps of the event recognition method described above.

Embodiments of the present application provide a storage medium for storing computer-executable instructions, which, when executed, implement the steps of the above-mentioned data tag generation method, or implement the steps of the above-mentioned model training method, or implement the steps of the above-mentioned event recognition method.

In this embodiment, first, a plurality of pieces of event data of a target event are obtained, and each piece of event data is processed through a plurality of event identification strategies corresponding to the target event, so as to obtain first identification result data of each piece of event data relative to a risk event, then, according to a data inference model corresponding to the first identification result data, data inference is performed based on the first identification result data, so as to obtain second identification result data of each piece of event data relative to the risk event, and finally, according to the second identification result data of each piece of event data, a data tag of each piece of event data is determined. In the embodiment, the identification result data of each event data can be accurately obtained through a data inference mode, so that the data label of the event data is accurately determined, the problem that label accuracy is difficult to guarantee in manual label printing is solved, the data label printing accuracy is improved, and the accuracy of a data processing result of a model is improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only some embodiments described in the present application, and for those skilled in the art, other drawings can be obtained according to the drawings without any creative effort.

Fig. 1 is a schematic view of an application scenario of each method in this embodiment according to an embodiment of the present application;

fig. 2 is a schematic flowchart of a data tag generation method according to an embodiment of the present application;

fig. 3 is a schematic flowchart of a data tag generation method according to another embodiment of the present application;

fig. 4 is a schematic flowchart of a data tag generation method according to another embodiment of the present application;

FIG. 5 is a schematic flow chart illustrating a model training method according to an embodiment of the present disclosure;

fig. 6 is a schematic flowchart of an event identification method according to an embodiment of the present application;

fig. 7 is a schematic diagram illustrating a module composition of a data tag generation apparatus according to an embodiment of the present application;

FIG. 8 is a block diagram of a model training apparatus according to an embodiment of the present disclosure;

fig. 9 is a schematic block diagram illustrating an event recognition apparatus according to an embodiment of the present application;

fig. 10 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

In order to make those skilled in the art better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The embodiment of the application provides a data tag generation method, a model training method and an event recognition device, and aims to solve the problem that the tag accuracy is difficult to guarantee in manual tagging and improve the accuracy of a data processing result of a model.

Fig. 1 is a schematic view of an application scenario of each method in this embodiment provided in an embodiment of the present application, and as shown in fig. 1, fig. 1 provides a computer system including a data tag generation subsystem 100, a model training subsystem 200, and an event identification subsystem 300, where the data tag generation subsystem 100 may execute the data tag generation method in this embodiment to tag each piece of event data. The model training subsystem 200 may execute the model training method in this embodiment, and train a risk event recognition model based on each piece of event data after tagging, where the risk event recognition model is used to recognize whether a target event is a risk event, or is used to recognize a probability that the target event is a risk event. The event recognition subsystem 300 may execute the event recognition method in this embodiment, and run the risk event recognition model trained by the model training subsystem 200, so as to recognize whether the target event is a risk event or not, or recognize the probability that the target event is a risk event.

In this embodiment, the data tag generation subsystem 100, the model training subsystem 200, and the event recognition subsystem 300 in fig. 1 may respectively include one or more computer devices, for example, the data tag generation subsystem 100 is composed of a plurality of computer devices, the model training subsystem 200 is composed of a plurality of computer devices, and the event recognition subsystem 300 is composed of a plurality of computer devices.

In other embodiments, the methods in this embodiment may be implemented by two subsystems, for example, the data label generating method and the model training method are executed by one subsystem, and the event recognition method is executed by the other subsystem, or the data label generating method is executed by one subsystem and the model training method and the event recognition method are executed by the other subsystem. Wherein each subsystem comprises one or more computer devices, respectively.

In other embodiments, the data label generation method, the model training method and the event recognition method can be implemented by one computer device. It should be noted that the functions and effects achieved by the above-mentioned various embodiments are the same, and are not repeated here.

Fig. 2 is a schematic flowchart of a data tag generation method according to an embodiment of the present application, and as shown in fig. 2, the flowchart includes the following steps:

step S202, acquiring a plurality of event data of a target event;

step S204, processing each event data respectively through a plurality of event identification strategies corresponding to the target event to obtain first identification result data of each event data relative to the risk event;

step S206, according to the data inference model corresponding to the first identification result data, performing data inference based on the first identification result data to obtain second identification result data of each event data relative to the risk event;

step S208, determining the data label of each event data according to the second identification result data of each event data.

In the step S202, the target event may be a user transaction event, and the event data of the target event may be transaction data corresponding to the user transaction event. For example, when a user performs a plurality of commodity purchasing activities through the e-commerce platform, the target event is set as a trading transaction event of the user, and the event data of the target event is set as transaction data corresponding to the trading transaction event of the user. In this embodiment, the transaction data includes, but is not limited to, transaction time, transaction amount, network environment parameters during transaction, device parameters of a terminal device used for transaction, a transaction amount deduction route, merchant information, collection account information, and the like.

In a specific embodiment, a credit card transaction event of a user is selected as a target event, and a risk event is set as a cash-out event, so that a label is marked for credit card transaction data of the user, and the label can be used for indicating the probability that the corresponding credit card transaction event is the cash-out event. In another embodiment, a user's shopping transaction event is selected as a target event and a risk event is set as a gambling event, thereby tagging the user's shopping transaction data with a tag that may be used to indicate a probability that the corresponding shopping transaction event is a gambling event.

In this embodiment, a risk data warehouse is preset, transaction data of a large number of users are stored in the risk data warehouse, and in step S202, a plurality of transaction data may be acquired from the risk data warehouse as event data.

In this embodiment, a plurality of event identification policies are pre-configured, and the event identification policies may process event data of a target event to identify whether the target event is a risk event or identify a probability that the target event is a risk event. Taking the target event as a credit card transaction event and the risk event as a cash-out event as an example, whether the credit card transaction event is the cash-out event can be identified through the event identification strategy. The event recognition policy may include various models or rules, etc.

In step S204, each piece of event data is processed through a plurality of event identification policies corresponding to the target event, so as to obtain first identification result data of each piece of event data with respect to the risk event. Since each piece of event data is processed by the plurality of event recognition policies, each piece of event data has a plurality of pieces of first recognition result data. In one embodiment, each piece of event data is identified by one event identification policy to obtain one piece of first identification result data, and the number of the first identification result data of each piece of event data is the same as that of the event identification policies.

In a particular embodiment, the event recognition policy includes a model and a rule. Event data can be processed by the model to determine the probability that a target event is a risk event, and event data can be processed by a rule (such as If-Then based rule) to determine whether the target event is a risk event. Wherein the respective thresholds involved in the models and rules may be determined based on manual experience. In this embodiment, if the output result of the model or rule is in the form of scores, normalization processing may be performed on each score output by the model or rule, for example, the scores are all normalized to be between 0 and 10 minutes, so that subsequent processing is facilitated.

Since the first recognition result data obtained by the event recognition policy may be inaccurate, and there may be a problem of low accuracy in tagging the event data based on the first recognition result data, the first recognition result data may also be referred to as a "weak tag" of the event data in this embodiment.

Table 1 below is a schematic table of first identification result data provided in an embodiment of the present application, and as shown in table 1, three event identification policies and four event data are taken as an example, and by using each event identification policy, one first identification result data can be obtained for each event data through identification, so as to obtain table 1 below.

TABLE 1

After obtaining the plurality of first recognition result data, in this embodiment, the data inference model corresponding to the first recognition result data may be determined in the following specific manner:

(a1) determining a plurality of first recognition result data of each event recognition strategy aiming at each event data as a group of first recognition result data before obtaining second recognition result data of each event data relative to the risk event;

(a2) analyzing each group of first recognition result data to determine a data correlation analysis result between each group of first recognition result data;

(a3) respectively carrying out distribution statistics on each group of first identification result data to determine distribution data corresponding to each group of first identification result data; wherein, the first identification result data can be score data;

(a4) and determining a data inference model corresponding to the first recognition result data according to the data correlation analysis result and/or the distribution data, wherein the data inference model comprises at least one of a voting model, a probability map model and a matrix decomposition model.

As can be seen from table 1, each event identification policy corresponds to a set of first identification result data based on a plurality of pieces of event data, for example, the first identification result data 1.1, the first identification result data 1.2, the first identification result data 1.3, and the first identification result data 1.4 in table 1 form a set of first identification result data. Therefore, in the above-mentioned act (a1), before obtaining the second recognition result data of each event data with respect to the risk event, the plurality of first recognition result data for each event data of each event recognition policy are determined as one set of first recognition result data, and there are three sets of first recognition result data in table 1.

In the above-mentioned act (a2), the sets of first recognition result data are analyzed to determine the data correlation analysis result between the sets of first recognition result data, for example, each column of first recognition result data from left to right in table 1 is determined as one set of first recognition result data, and the three sets of first recognition result data in table 1 are analyzed to determine the data correlation between any two sets of first recognition result data. The data correlation means that there is a certain relationship between data, such as positive correlation, negative correlation, etc., and the specific analysis method may adopt a general analysis algorithm, which is not limited herein.

In the above-described operation (a3), distribution statistics is performed on each of the sets of first recognition result data to determine distribution data corresponding to each of the sets of first recognition result data. In a specific embodiment, the distribution statistics is performed on each group of the first recognition result data to obtain a data distribution function corresponding to each group of the first recognition result data, and the data distribution function is used as the distribution data corresponding to each group of the first recognition result data. Of course, the distribution data may be a data distribution histogram, a data distribution graph, or the like, and is not limited herein.

In the above-mentioned act (a4), a data inference model corresponding to the first recognition result data is determined based on the data correlation analysis result and/or the distribution data, and the data inference model includes at least one of a voting model, a probability map model, and a matrix decomposition model.

In one embodiment, a model may be manually selected from the voting model, the probabilistic graphical model, and the matrix factorization model according to the data correlation analysis result and/or the distribution data, and is used as the data inference model corresponding to the first recognition result data.

In another embodiment, if the data correlation analysis result indicates a high correlation between the sets of first recognition result data, a probabilistic graph model or a matrix decomposition model may be selected as the data inference model, and if the data correlation analysis result indicates a no correlation between the sets of first recognition result data, a voting model may be selected as the data inference model.

In another embodiment, a probabilistic graph model or a matrix decomposition model may be selected as the data estimation model if the distribution data satisfies a specific distribution form requirement, and a voting model may be selected as the data estimation model if the distribution data does not satisfy the specific distribution form requirement.

In another embodiment, if the data correlation analysis result indicates a high correlation between the sets of first recognition result data and the distribution data satisfies a specific distribution form requirement, a probabilistic graph model or a matrix decomposition model may be selected as the data inference model, and if the data correlation analysis result indicates a correlation between the sets of first recognition result data or the distribution data does not satisfy the specific distribution form requirement, a voting model may be selected as the data inference model.

There are various ways to select one model from the voting model, the probabilistic graphical model and the matrix factorization model as the data inference model according to the data correlation analysis result and/or the distribution data, and this is not listed here.

After determining the data inference model, step S206 may be executed to perform data inference based on the first recognition result data according to the data inference model corresponding to the first recognition result data, so as to obtain second recognition result data of each event data with respect to the risk event, where the process specifically includes: and calculating each first recognition result data corresponding to each event data through the data inference model to obtain second recognition result data of each event data relative to the risk event.

Specifically, taking the data inference model as the voting model as an example, and combining table 1, in this step, the voting model may perform voting operation on each piece of first recognition result data corresponding to each piece of event data to obtain second recognition result data of each piece of event data relative to the risk event. The second recognition result data may be determined by voting selection from the respective first recognition result data, or may be newly generated recognition result data.

Taking the data inference model as the probability map model as an example, and combining the table 1, in this step, a probability map may be established by the probability map model based on each first recognition result data corresponding to each event data, and second recognition result data of each event data relative to the risk event may be determined according to the probability map. The second recognition result data may be selected from the first recognition result data, or may be newly generated recognition result data.

The matrix decomposition model is used for calculating each first recognition result data corresponding to each event data, and the process of obtaining the second recognition result data of each event data relative to the risk event can refer to the specific algorithm process of the matrix decomposition model, and is not repeated here.

In this embodiment, each first recognition result data corresponding to each event data is calculated in a data inference manner to obtain a second recognition result data of each event data relative to the risk event, so that the obtained second recognition result data has the advantage of high accuracy.

In step S208, determining the data tag of each event data according to the second identification result data of each event data may be: the second recognition result data of each event data is respectively determined as the data label of each event data, for example, the data label of the first event data is 1, which indicates that the target event corresponding to the event data is a gambling event, and the data label of the second event data is 0, which indicates that the target event corresponding to the event data is not a gambling event, wherein the gambling event is the aforementioned risk event.

Through the process, accurate data labels can be determined for all event data, so that the problem that label accuracy is difficult to guarantee in manual label printing is solved, the label printing accuracy of data is improved, and the accuracy of a data processing result of a model is improved.

Considering that after the first recognition result data of each piece of event data relative to the risk event is obtained in step S204, there may be a case that the first recognition result data is missing data, or the first recognition result data is too inaccurate and does not have reference, the method in this embodiment may further, before obtaining the second recognition result data of each piece of event data relative to the risk event: and determining a problem strategy in each event identification strategy according to the first identification result data, and deleting the first identification result data corresponding to the problem strategy in the first identification result data. Correspondingly, according to the data inference model corresponding to the first recognition result data, performing data inference based on the first recognition result data, specifically: and performing data inference based on the deleted first recognition result data according to the data inference model corresponding to the deleted first recognition result data.

According to the first identification result data, determining a problem policy in each event identification policy may be:

(b1) determining a plurality of first recognition result data for each event recognition strategy for each piece of event data as a set of first recognition result data;

(b2) executing one or more of the following processes on the first identification result data respectively, and determining a problem strategy in each event identification strategy according to the execution result;

(b21) respectively counting the quantity proportion of first identification result data which represent the identification result as vacancy in each group of first identification result data;

(b22) respectively carrying out distribution statistics on each group of first identification result data to determine distribution data corresponding to each group of first identification result data; wherein, the first identification result data can be score data;

(b23) and analyzing the groups of first recognition result data to determine data correlation analysis results among the groups of first recognition result data.

As can be seen from table 1, each event identification policy corresponds to a set of first identification result data based on a plurality of pieces of event data, for example, the first identification result data 1.1, the first identification result data 1.2, the first identification result data 1.3, and the first identification result data 1.4 in table 1 form a set of first identification result data. Therefore, in the above-mentioned action (b1), the plurality of first recognition result data for each event data of each event recognition policy is determined as one set of first recognition result data, and three sets of first recognition result data are common in table 1.

In the above operation (b21), the number ratio of the first recognition result data indicating that the recognition result is empty among the respective sets of the first recognition result data is counted. Taking table 1 as an example, the number ratio of the first recognition result data indicating that the recognition result is a vacancy in each column of data (i.e., each group of the first recognition result data) in table 1 is counted. When the event data of the target event is processed through the event identification policy, the identification result data of the event identification policy may indicate that the identification result is empty, that is, the identification result is not obtained.

In the above-described operation (b22), distribution statistics is performed on each of the groups of first recognition result data to determine distribution data corresponding to each of the groups of first recognition result data. In a specific embodiment, the distribution statistics is performed on each group of the first recognition result data to obtain a data distribution function corresponding to each group of the first recognition result data, and the data distribution function is used as the distribution data corresponding to each group of the first recognition result data. Of course, the distribution data may be a data distribution histogram, a data distribution graph, or the like, and is not limited herein.

In the above-mentioned act (b23), the sets of first recognition result data are analyzed to determine the data correlation analysis result between the sets of first recognition result data, for example, each column of first recognition result data from left to right in table 1 is determined as one set of first recognition result data, the three sets of first recognition result data in table 1 are analyzed to determine the data correlation between any two sets of first recognition result data, or the data correlation between any three sets of first recognition result data is determined. The data correlation means that there is a certain relationship between data, such as positive correlation, negative correlation, etc., and the specific analysis method may adopt a general analysis algorithm, which is not limited herein.

In the above-described operation (b2), the problem policy is determined among the event recognition policies based on the execution result. For example, one or more sets of first recognition result data indicating that the ratio of the number of first recognition result data indicating that the recognition result is a vacancy exceeds a set ratio threshold are determined from the ratio of the number counted in the operation (b21), and the event recognition policy corresponding to the one or more sets of first recognition result data is determined as the problem policy. For another example, according to the distribution data determined in the action (b22), one or more sets of first recognition result data whose distribution form of the corresponding distribution data does not meet the preset distribution form requirement are determined, and the event recognition policy corresponding to the one or more sets of first recognition result data is determined as the problem policy. For another example, two sets of highly correlated first recognition result data are determined from the data correlation analysis result determined in the action (b23), and the event recognition policy corresponding to any one set of first recognition result data in the two sets of first recognition result data is determined as the problem policy.

In a specific embodiment, if it is determined from the data correlation analysis result that there are two sets of first recognition result data whose data correlations are highly correlated, and the quantity ratio of the first recognition result data indicating that the recognition result is a vacancy in one set of first recognition result data is greater than a preset ratio threshold, the event recognition policy corresponding to any one set of first recognition result data in the two sets of first recognition result data is determined as the problem policy.

In another specific embodiment, the vacancy value proportion requirement, the distributed data requirement, the data correlation requirement, and the like may be determined, the problem policy may be determined among the event recognition policies according to the requirements and the execution result of the above-mentioned actions (b21) (b22) (b23), then the first recognition result data corresponding to the problem policy may be deleted from the first recognition result data, and the data inference model corresponding to the deleted first recognition result data may be determined.

In yet another embodiment, a data inference model corresponding to the first recognition result data may be determined, and then a corresponding data requirement including a vacancy value proportion requirement, a distributed data requirement, a data correlation requirement, etc. may be determined according to the performance parameters of the data inference model, and a problem policy may be determined among the event recognition policies according to the requirements and the execution results of the above actions (b21) (b22) (b 23). In this embodiment, the data inference model corresponding to the deleted first recognition result data is a data inference model corresponding to the first recognition result data before deletion.

Fig. 3 is a schematic flowchart of a data tag generation method according to another embodiment of the present application, and as shown in fig. 3, the flowchart includes the following steps:

step S302, acquiring a plurality of event data of a target event;

step S304, processing each event data respectively through a plurality of event identification strategies corresponding to the target event to obtain first identification result data of each event data relative to the risk event;

step S306, according to the first identification result data, determining a problem strategy in each event identification strategy, and deleting the first identification result data corresponding to the problem strategy in the first identification result data;

step S308, determining a data inference model corresponding to the deleted first recognition result data;

step S310, according to the data inference model corresponding to the deleted first identification result data, performing data inference based on the deleted first identification result data to obtain second identification result data of each event data relative to the risk event;

in step S312, the second identification result data of each piece of event data is determined as the data tag of each piece of event data.

Fig. 4 is a schematic flowchart of a data tag generation method according to another embodiment of the present application, and as shown in fig. 4, the flowchart includes the following steps:

step S402, acquiring a plurality of event data of a target event;

step S404, processing each event data respectively through a plurality of event identification strategies corresponding to the target event to obtain first identification result data of each event data relative to the risk event;

step S406, determining a data inference model corresponding to the first recognition result data;

step S408, according to the first identification result data and the data requirement corresponding to the data inference model, determining a problem strategy in each event identification strategy, and deleting the first identification result data corresponding to the problem strategy in the first identification result data;

step S410, according to a data inference model corresponding to the first identification result data, performing data inference based on the deleted first identification result data to obtain second identification result data of each event data relative to the risk event;

in step S412, the second identification result data of each piece of event data is determined as the data tag of each piece of event data.

Through the process shown in fig. 3 and fig. 4, accurate data labels can be determined for each piece of event data, so that the problem that label accuracy is difficult to guarantee in manual label printing is solved, the data label printing accuracy is improved, and the accuracy of a data processing result of a model is improved. In particular, through the flow in fig. 4, the data that does not meet the requirement can be determined in the first recognition result data according to the data requirement corresponding to the data inference model, and deleted, so that the data tag inference result can be more accurate.

The data tag generation method in the embodiment can be applied in the field of fund risk identification, and can achieve the effect of tagging event data by setting a target event as a transaction event of a user, and setting event data of the target event as event data of the transaction event of the user, wherein the risk event can be a money laundering event, a gambling event, a cash register event and the like.

In the fund risk identification field, the method in the embodiment is used for labeling the transaction data, so that the labeling accuracy can be improved, the fund risk identification accuracy of a user can be improved, and the method can be automatically executed without manual intervention, so that the method is simple and convenient to operate and maintain.

In summary, the data tag generation method in the embodiment has the following beneficial effects:

(1) accurate data labels can be determined for each event data, so that the problem that label accuracy is difficult to guarantee in manual label printing is solved, the label printing accuracy of the data is improved, and the accuracy of a data processing result of a model is improved;

(2) the method is characterized in that a data label generation method based on weak supervised learning is provided, the problems of low accuracy and low coverage rate of the existing identification model and rule are improved, the problem of essentially unsupervised learning is started from label learning, and a solution thought and a solution way based on a weak label are provided;

(3) the method can be applied to a capital risk identification scene, and accuracy of user capital risk identification is improved;

(4) the method can be realized by using various computer languages and software and hardware, and is not limited by software and hardware environments and computer languages.

Based on the above data tag generation method, an embodiment of the present application further provides a model training method, fig. 5 is a schematic flow chart of the model training method provided in the embodiment of the present application, and as shown in fig. 5, the method includes the following steps:

step S502, determining the data label of the event data of the target event according to the data label generating method;

reference is made to the preceding description for this part, which is not repeated here.

Step S504, taking the event data of the target event as sample data of the target event;

step S506, training a risk event identification model corresponding to the target event according to the sample data of the target event and the data label of the sample data, wherein the risk event identification model is used for identifying whether the target event is a risk event or not, or identifying the probability that the target event is a risk event.

The model in step S506 may be trained by using an existing method, which is not limited herein. The risk event recognition model obtained by training in this embodiment may be a deep learning model or a convolutional neural network model.

In this embodiment, the target event may be a user transaction event, and the event data of the target event may be transaction data corresponding to the user transaction event. For example, when a user performs a plurality of commodity purchasing activities through the e-commerce platform, the target event is set as a trading transaction event of the user, and the event data of the target event is set as transaction data corresponding to the trading transaction event of the user. In this embodiment, the transaction data includes, but is not limited to, transaction time, transaction amount, network environment parameters during transaction, device parameters of a terminal device used for transaction, a transaction amount deduction route, merchant information, collection account information, and the like.

In a specific embodiment, a credit card transaction event of a user is selected as a target event, a risk event is set as a cash-out event, so that credit card transaction data of the user is labeled, and a risk event identification model for identifying the cash-out event is trained based on the labeled risk event. In another embodiment, a user's shopping transaction event is selected as a target event and a risk event is set as a gambling event, such that the user's shopping transaction data is tagged and a risk event recognition model for recognizing the gambling event is trained based on the tagged tags.

In this embodiment, the risk event recognition model is trained by using the data labels generated by the data label generation method. Because the identification result data of each event data can be accurately obtained in a data inference mode when the data label is generated, the data label of the event data can be accurately determined, the problem that the label accuracy is difficult to guarantee in manual label printing is solved, the label printing accuracy of the data is improved, the model training accuracy is improved, and the accuracy of the data processing result of the model is improved.

Based on the above model training method, an embodiment of the present application further provides an event recognition method, fig. 6 is a schematic flow chart of the event recognition method provided in the embodiment of the present application, and as shown in fig. 6, the method includes the following steps:

step S602, acquiring event data of a target event to be identified.

Specifically, event data of the target event to be identified may be acquired from the network.

And step S604, processing the event data of the target event to be recognized by utilizing the risk event recognition model obtained by training through the model training method.

Step S606, according to the processing result, determining whether the target event to be identified is a risk event, or determining the probability that the target event to be identified is a risk event.

The output result of the risk event identification model may be whether the target event to be identified is a risk event or not, or the probability that the target event to be identified is a risk event.

In this embodiment, the target event to be identified may be a user transaction event, and the event data of the target event to be identified may be transaction data corresponding to the user transaction event. For example, when a user performs multiple commodity purchasing activities through the e-commerce platform, the target event to be identified is set as a trading transaction event of the user, and event data of the target event to be identified is set as transaction data corresponding to the trading transaction event of the user. In this embodiment, the transaction data includes, but is not limited to, transaction time, transaction amount, network environment parameters during transaction, device parameters of a terminal device used for transaction, a transaction amount deduction route, merchant information, collection account information, and the like.

In a specific embodiment, the credit card transaction event of the user is selected as the target event to be identified, and the risk event is set as the cash register event, so that whether the credit card transaction event is the cash register event or not is identified. In another embodiment, a shopping transaction event of a user is selected as a target event to be identified, and a risk event is set as a gambling event, thereby identifying whether the shopping transaction event is a gambling event.

The processing result of the risk event recognition model in this embodiment may be a two-classification result or a score result.

In this embodiment, the risk event recognition model obtained by training with the model training method identifies whether the target event to be recognized is a risk event. When the risk event recognition model is trained, the recognition result data of each event data is accurately obtained by adopting the data label generation method in a data inference mode, so that the data label of the event data can be accurately determined, the data labeling accuracy is improved, the model training accuracy is improved, and the accuracy of the data processing result of the model is improved.

Based on the foregoing data tag generation method, an embodiment of the present application further provides a data tag generation apparatus, and fig. 7 is a schematic diagram of module composition of the data tag generation apparatus provided in the embodiment of the present application, and as shown in fig. 7, the apparatus includes:

a first acquiring module 71, configured to acquire a plurality of pieces of event data of a target event; a first result determining module 72, configured to respectively process each piece of event data through a plurality of event identification policies corresponding to the target event, so as to obtain first identification result data of each piece of event data relative to a risk event; a second result determining module 73, configured to perform data inference based on the first recognition result data according to a data inference model corresponding to the first recognition result data, to obtain second recognition result data of each piece of event data relative to a risk event; and a tag generating module 74, configured to determine a data tag of each piece of event data according to the second identification result data of each piece of event data.

Optionally, the apparatus further comprises a model determination module configured to: determining a plurality of first recognition result data of each event recognition strategy for each piece of event data as a set of the first recognition result data before obtaining second recognition result data of each piece of event data relative to a risk event; analyzing each group of the first identification result data to determine a data correlation analysis result between each group of the first identification result data; respectively carrying out distribution statistics on each group of the first identification result data to determine distribution data corresponding to each group of the first identification result data; and determining a data inference model corresponding to the first identification result data according to the data correlation analysis result and/or the distribution data, wherein the data inference model comprises at least one of a voting model, a probability map model and a matrix decomposition model.

Optionally, the apparatus further includes a data deleting module, configured to: before second identification result data of each event data relative to the risk event is obtained, determining a problem strategy in each event identification strategy according to the first identification result data, and deleting first identification result data corresponding to the problem strategy in the first identification result data; the second result determining module 73 is specifically configured to: and performing data inference based on the deleted first recognition result data according to the data inference model corresponding to the deleted first recognition result data.

Optionally, the data deleting module is specifically configured to: determining a plurality of the first recognition result data for each of the event data of each of the event recognition policies as a set of the first recognition result data; executing one or more of the following processes on the first identification result data respectively, and determining a problem strategy in each event identification strategy according to an execution result; respectively counting the quantity proportion of first identification result data which represent that identification results are vacant in each group of first identification result data; respectively carrying out distribution statistics on each group of the first identification result data to determine distribution data corresponding to each group of the first identification result data; and analyzing each group of the first recognition result data to determine a data correlation analysis result between each group of the first recognition result data.

Optionally, the data inference model comprises at least one of a voting model, a probabilistic graphical model, and a matrix decomposition model; the second result determining module 73 is specifically configured to: and calculating each first recognition result data corresponding to each event data through the data inference model to obtain second recognition result data of each event data relative to the risk event.

Optionally, the tag generating module 74 is specifically configured to: and respectively determining the second identification result data of each piece of event data as a data tag of each piece of event data.

Based on the above model training method, an embodiment of the present application further provides a model training device, fig. 8 is a schematic diagram of a module composition of the model training device provided in an embodiment of the present application, and as shown in fig. 8, the device includes:

a tag determination module 81, configured to determine a data tag of event data of a target event according to the data tag generation apparatus;

a sample determining module 82, configured to use the event data of the target event as sample data of the target event;

and a model training module 83, configured to train a risk event identification model corresponding to the target event according to the sample data of the target event and the data tag of the sample data, where the risk event identification model is used to identify whether the target event is a risk event, or is used to identify a probability that the target event is a risk event.

In this embodiment, the risk event recognition model is trained using the data labels generated by the data label generation apparatus. Because the identification result data of each event data can be accurately obtained in a data inference mode when the data label is generated, the data label of the event data can be accurately determined, the problem that the label accuracy is difficult to guarantee in manual label printing is solved, the label printing accuracy of the data is improved, the model training accuracy is improved, and the accuracy of the data processing result of the model is improved.

Based on the above event identification method, an embodiment of the present application further provides an event identification device, and fig. 9 is a schematic diagram of module composition of the event identification device provided in the embodiment of the present application, as shown in fig. 9, the device includes:

a second obtaining module 91, configured to obtain event data of a target event to be identified;

the data processing module 92 is configured to process event data of the target event to be recognized by using a risk event recognition model obtained through training by the model training device;

and the event identification module 93 is configured to determine, according to the processing result, whether the target event to be identified is a risk event, or determine a probability that the target event to be identified is a risk event.

In this embodiment, whether the target event to be identified is a risk event is identified by using the risk event identification model obtained by training with the model training device. When the risk event recognition model is trained, the data label generation device is adopted to accurately obtain the recognition result data of each event data in a data inference mode, so that the data label of the event data can be accurately determined, the data labeling accuracy is improved, the model training accuracy is improved, and the accuracy of the data processing result of the model is improved.

It should be noted that the data tag generation apparatus, the model training apparatus, and the event recognition apparatus in the embodiment of the present application may respectively implement each process of the embodiments of the foregoing data tag generation method, model training method, and event recognition method, and achieve the same effect and function, which is not described herein again.

Further, an embodiment of the present application also provides an electronic device, and fig. 10 is a schematic structural diagram of the electronic device provided in the embodiment of the present application, as shown in fig. 10. Electronic devices may vary widely in configuration or performance and may include one or more processors 901 and memory 902, where the memory 902 may store one or more stored applications or data. Memory 902 may be, among other things, transient storage or persistent storage. The application program stored in memory 902 may include one or more modules (not shown), each of which may include a series of computer-executable instructions for the electronic device. Still further, the processor 901 may be configured to communicate with the memory 902 to execute a series of computer-executable instructions in the memory 902 on the electronic device. The electronic device may also include one or more power supplies 903, one or more wired or wireless network interfaces 904, one or more input-output interfaces 905, one or more keyboards 906, and the like.

In one particular embodiment, an electronic device includes memory, and one or more programs, where the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer-executable instructions for the electronic device, and execution of the one or more programs by one or more processors includes computer-executable instructions for:

acquiring a plurality of pieces of event data of a target event;

Optionally, the computer executable instructions, when executed, further comprise, before obtaining second identification result data of each of the event data with respect to the risk event: determining a plurality of the first recognition result data for each of the event data of each of the event recognition policies as a set of the first recognition result data; analyzing each group of the first identification result data to determine a data correlation analysis result between each group of the first identification result data; respectively carrying out distribution statistics on each group of the first identification result data to determine distribution data corresponding to each group of the first identification result data; and determining a data inference model corresponding to the first identification result data according to the data correlation analysis result and/or the distribution data, wherein the data inference model comprises at least one of a voting model, a probability map model and a matrix decomposition model.

Optionally, the computer executable instructions, when executed, further comprise, before obtaining second identification result data of each of the event data with respect to the risk event: according to the first identification result data, determining a problem strategy in each event identification strategy, and deleting the first identification result data corresponding to the problem strategy in the first identification result data; performing data inference based on the first recognition result data according to a data inference model corresponding to the first recognition result data, including: and performing data inference based on the deleted first recognition result data according to the data inference model corresponding to the deleted first recognition result data.

Optionally, when executed, the computer-executable instructions determine a problem policy among the event identification policies according to the first identification result data, including: determining a plurality of the first recognition result data for each of the event data of each of the event recognition policies as a set of the first recognition result data; executing one or more of the following processes on the first identification result data respectively, and determining a problem strategy in each event identification strategy according to an execution result; respectively counting the quantity proportion of first identification result data which represent that identification results are vacant in each group of first identification result data; respectively carrying out distribution statistics on each group of the first identification result data to determine distribution data corresponding to each group of the first identification result data; and analyzing each group of the first recognition result data to determine a data correlation analysis result between each group of the first recognition result data.

Optionally, the computer-executable instructions, when executed, the data inference model comprises at least one of a voting model, a probabilistic graphical model, and a matrix decomposition model; according to a data inference model corresponding to the first recognition result data, performing data inference based on the first recognition result data to obtain second recognition result data of each event data relative to a risk event, wherein the data inference model comprises: and calculating each first recognition result data corresponding to each event data through the data inference model to obtain second recognition result data of each event data relative to the risk event.

Optionally, when executed, the computer-executable instructions determine a data tag of each piece of event data according to the second recognition result data of each piece of event data, including: and respectively determining the second identification result data of each piece of event data as a data tag of each piece of event data.

In another particular embodiment, an electronic device includes a memory, and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer-executable instructions for the electronic device, and the one or more programs configured for execution by the one or more processors include computer-executable instructions for:

taking the event data of the target event as sample data of the target event;

acquiring event data of a target event to be identified;

It should be noted that the electronic device in each embodiment of the present application may respectively implement each process of the embodiments of the foregoing data tag generation method, model training method, and event recognition method, and achieve the same effect and function, which is not described herein again.

Further, embodiments of the present application also provide a storage medium for storing computer-executable instructions, in a specific embodiment, the storage medium may be a usb disk, an optical disk, a hard disk, and the like, and the storage medium stores computer-executable instructions that, when executed by a processor, implement the following processes:

acquiring a plurality of pieces of event data of a target event;

Optionally, the storage medium stores computer-executable instructions that, when executed by the processor, further comprise, before obtaining second identification result data of each of the event data with respect to the risk event: determining a plurality of the first recognition result data for each of the event data of each of the event recognition policies as a set of the first recognition result data; analyzing each group of the first identification result data to determine a data correlation analysis result between each group of the first identification result data; respectively carrying out distribution statistics on each group of the first identification result data to determine distribution data corresponding to each group of the first identification result data; and determining a data inference model corresponding to the first identification result data according to the data correlation analysis result and/or the distribution data, wherein the data inference model comprises at least one of a voting model, a probability map model and a matrix decomposition model.

Optionally, the storage medium stores computer-executable instructions that, when executed by the processor, further comprise, before obtaining second identification result data of each of the event data with respect to the risk event: according to the first identification result data, determining a problem strategy in each event identification strategy, and deleting the first identification result data corresponding to the problem strategy in the first identification result data; performing data inference based on the first recognition result data according to a data inference model corresponding to the first recognition result data, including: and performing data inference based on the deleted first recognition result data according to the data inference model corresponding to the deleted first recognition result data.

Optionally, the storage medium stores computer-executable instructions that, when executed by the processor, determine a problem policy among the event identification policies based on the first identification result data, including: determining a plurality of the first recognition result data for each of the event data of each of the event recognition policies as a set of the first recognition result data; executing one or more of the following processes on the first identification result data respectively, and determining a problem strategy in each event identification strategy according to an execution result; respectively counting the quantity proportion of first identification result data which represent that identification results are vacant in each group of first identification result data; respectively carrying out distribution statistics on each group of the first identification result data to determine distribution data corresponding to each group of the first identification result data; and analyzing each group of the first recognition result data to determine a data correlation analysis result between each group of the first recognition result data.

Optionally, the storage medium stores computer-executable instructions that, when executed by the processor, cause the data inference model to include at least one of a voting model, a probabilistic graphical model, and a matrix factorization model; according to a data inference model corresponding to the first recognition result data, performing data inference based on the first recognition result data to obtain second recognition result data of each event data relative to a risk event, wherein the data inference model comprises: and calculating each first recognition result data corresponding to each event data through the data inference model to obtain second recognition result data of each event data relative to the risk event.

Optionally, the storage medium stores computer-executable instructions that, when executed by the processor, determine a data tag of each piece of event data according to the second recognition result data of each piece of event data, including: and respectively determining the second identification result data of each piece of event data as a data tag of each piece of event data.

In another specific embodiment, the storage medium may be a usb disk, an optical disk, a hard disk, or the like, and the storage medium stores computer executable instructions that, when executed by the processor, implement the following process:

taking the event data of the target event as sample data of the target event;

acquiring event data of a target event to be identified;

It should be noted that the storage medium in each embodiment of the present application may respectively implement each process of the embodiments of the foregoing data tag generation method, model training method, and event recognition method, and achieve the same effect and function, which is not described herein again.

In the 90 s of the 20 th century, improvements in a technology could clearly distinguish between improvements in hardware (e.g., improvements in circuit structures such as diodes, transistors, switches, etc.) and improvements in software (improvements in process flow). However, as technology advances, many of today's process flow improvements have been seen as direct improvements in hardware circuit architecture. Designers almost always obtain the corresponding hardware circuit structure by programming an improved method flow into the hardware circuit. Thus, it cannot be said that an improvement in the process flow cannot be realized by hardware physical modules. For example, a Programmable Logic Device (PLD), such as a Field Programmable Gate Array (FPGA), is an integrated circuit whose Logic functions are determined by programming the Device by a user. A digital system is "integrated" on a PLD by the designer's own programming without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing an integrated Circuit chip, such Programming is often implemented by "logic compiler" software, which is similar to a software compiler used in program development and writing, but the original code before compiling is also written by a specific Programming Language, which is called Hardware Description Language (HDL), and HDL is not only one but many, such as ABEL (Advanced Boolean Expression Language), AHDL (alternate Hardware Description Language), traffic, CUPL (core universal Programming Language), HDCal, jhddl (Java Hardware Description Language), Lava, Lola, HDL, PALASM, rhyd (Hardware Description Language), and the like, which are currently used in the field-Hardware Language. It will also be apparent to those skilled in the art that hardware circuitry that implements the logical method flows can be readily obtained by merely slightly programming the method flows into an integrated circuit using the hardware description languages described above.

The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic for the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller as pure computer readable program code, the same functionality can be implemented by logically programming method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Such a controller may thus be considered a hardware component, and the means included therein for performing the various functions may also be considered as a structure within the hardware component. Or even means for performing the functions may be regarded as being both a software module for performing the method and a structure within a hardware component.

The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. One typical implementation device is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functionality of the units may be implemented in one or more software and/or hardware when implementing the present application.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.

Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.

The above description is only an example of the present application and is not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A data tag generation method is characterized by comprising the following steps:

acquiring a plurality of pieces of event data of a target event;

respectively processing each piece of event data through a plurality of event identification strategies corresponding to the target event to obtain first identification result data of each piece of event data relative to a risk event, and determining the first identification result data of each event identification strategy aiming at each piece of event data as a group of first identification result data;

analyzing each group of the first identification result data to determine a data correlation analysis result between each group of the first identification result data; respectively carrying out distribution statistics on each group of the first identification result data to determine distribution data corresponding to each group of the first identification result data;

determining a data inference model corresponding to the first recognition result data according to the data correlation analysis result and/or the distribution data, wherein the data inference model comprises at least one of a voting model, a probability map model and a matrix decomposition model;

determining a data label of each piece of event data according to the second identification result data of each piece of event data; the data labels are used for training a risk event identification model corresponding to the target event, and the risk event identification model is used for identifying whether the target event is a risk event or not, or identifying the probability that the target event is a risk event.

2. The method of claim 1, wherein prior to obtaining second identification data for each of the event data relative to risk events, the method further comprises:

according to the first identification result data, determining a problem strategy in each event identification strategy, and deleting the first identification result data corresponding to the problem strategy in the first identification result data;

performing data inference based on the first recognition result data according to a data inference model corresponding to the first recognition result data, including:

and performing data inference based on the deleted first recognition result data according to the data inference model corresponding to the deleted first recognition result data.

3. The method of claim 2, wherein determining a problem policy among the event recognition policies based on the first recognition result data comprises:

determining a plurality of the first recognition result data for each of the event data of each of the event recognition policies as a set of the first recognition result data;

executing one or more of the following processes on the first identification result data respectively, and determining a problem strategy in each event identification strategy according to an execution result;

respectively counting the quantity proportion of first identification result data which represent that identification results are vacant in each group of first identification result data;

respectively carrying out distribution statistics on each group of the first identification result data to determine distribution data corresponding to each group of the first identification result data;

and analyzing each group of the first recognition result data to determine a data correlation analysis result between each group of the first recognition result data.

4. The method according to any one of claims 1 to 3, wherein performing data inference based on the first recognition result data according to a data inference model corresponding to the first recognition result data to obtain second recognition result data of each piece of event data relative to a risk event comprises:

and calculating each first recognition result data corresponding to each event data through the data inference model to obtain second recognition result data of each event data relative to the risk event.

5. The method according to any one of claims 1 to 3, wherein determining a data tag of each piece of event data based on the second recognition result data of each piece of event data includes:

and respectively determining the second identification result data of each piece of event data as a data tag of each piece of event data.

6. A method of model training, comprising:

the method according to any one of claims 1 to 5, determining a data tag of event data of a target event;

taking the event data of the target event as sample data of the target event;

7. An event recognition method, comprising:

acquiring event data of a target event to be identified;

processing the event data of the target event to be identified by utilizing the risk event identification model obtained by training in the claim 6;

8. A data tag generation apparatus, comprising:

a model determination module for determining a plurality of the first recognition result data for each of the event recognition policies for the respective pieces of the event data as a set of the first recognition result data; analyzing each group of the first identification result data to determine a data correlation analysis result between each group of the first identification result data; respectively carrying out distribution statistics on each group of the first identification result data to determine distribution data corresponding to each group of the first identification result data; determining a data inference model corresponding to the first recognition result data according to the data correlation analysis result and/or the distribution data, wherein the data inference model comprises at least one of a voting model, a probability map model and a matrix decomposition model;

a tag generation module, configured to determine a data tag of each piece of event data according to the second identification result data of each piece of event data; the data labels are used for training a risk event identification model corresponding to the target event, and the risk event identification model is used for identifying whether the target event is a risk event or not, or identifying the probability that the target event is a risk event.

9. The apparatus of claim 8, further comprising a data deletion module to:

before second identification result data of each event data relative to the risk event is obtained, determining a problem strategy in each event identification strategy according to the first identification result data, and deleting first identification result data corresponding to the problem strategy in the first identification result data;

the second result determination module is specifically configured to:

10. The apparatus of claim 9, wherein the data deletion module is specifically configured to:

11. The apparatus according to any one of claims 8 to 10, wherein the second result determination module is specifically configured to:

12. The apparatus according to any one of claims 8 to 10, wherein the tag generation module is specifically configured to:

13. A model training apparatus, comprising:

a tag determination module for determining a data tag of event data of a target event according to the apparatus of any one of claims 8 to 12;

14. An event recognition apparatus, comprising:

a data processing module, configured to process event data of the target event to be identified by using the risk event identification model obtained through training in claim 13;

15. An electronic device, comprising: a processor; and a memory arranged to store computer executable instructions which, when executed, cause the processor to carry out the steps of the data tag generation method of any one of the preceding claims 1 to 5, or the steps of the model training method of claim 6, or the steps of the event recognition method of claim 7.

16. A storage medium storing computer-executable instructions, wherein the computer-executable instructions, when executed, implement the steps of the data tag generation method of any one of claims 1 to 5, or implement the steps of the model training method of claim 6, or implement the steps of the event recognition method of claim 7.