CN112107866A

CN112107866A - User behavior data processing method, device, equipment and storage medium

Info

Publication number: CN112107866A
Application number: CN202011044215.5A
Authority: CN
Inventors: 刘志煌
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2020-09-28
Filing date: 2020-09-28
Publication date: 2020-12-22

Abstract

The embodiment of the application relates to cloud technology and artificial intelligence, and discloses a user behavior data processing method, a device, equipment and a storage medium, wherein the method comprises the following steps: acquiring historical mode sequence data and current mode sequence data of a target user; the historical pattern sequence data is obtained by processing based on historical behavior sequence data of the target user, and the current pattern sequence data is obtained by processing the current behavior sequence data of the target user; performing feature extraction processing on the historical pattern sequence data through a target feature extraction model to obtain historical pattern features, and performing feature extraction processing on the current pattern sequence data through the target feature extraction model to obtain current pattern features; and determining whether the current behavior corresponding to the current behavior sequence data of the target user is a cheating behavior according to the similarity between the historical pattern feature and the current pattern feature. The method can identify cheating behaviors with low cost, high applicability and high accuracy.

Description

User behavior data processing method, device, equipment and storage medium

Technical Field

The present application relates to the field of Artificial Intelligence (AI) technology and cloud technology, and in particular, to a method, an apparatus, a device, and a storage medium for processing user behavior data.

Background

Under the large environment that various scenes are vigorously developed by online and cloud technologies, more and more cheating behaviors are inevitably generated in a plurality of scenes (such as online competitions, advertisements, games and the like), and in order to better identify and attack the cheating behaviors, anti-cheating gradually becomes an important link of a plurality of online operation scenes in the control of a wind control security strategy.

The current common cheating behavior identification method mainly comprises the following steps: a method for discovering cheating behaviors based on a decision tree discrimination model and a method for identifying the cheating behaviors based on cheating rules set manually. In the method for discovering cheating behaviors based on the decision tree discrimination model, whether the cheating behaviors exist in the user can be identified by the decision tree discrimination model according to the characteristics of the user. In the method for identifying the cheating behaviors based on the cheating rules set manually, all measures possibly taken by the cheating need to be listed by combining the characteristics of specific application scenes, and then the cheating rules are combed based on the method, and the combed cheating rules are utilized to counter the cheating behaviors.

The cheating behavior identification method mainly has the following defects: the method for discovering cheating behaviors based on the decision tree discrimination model needs to train the decision tree discrimination model by using the labeled samples, the cost for constructing a large number of labeled samples is high, the model precision of the decision tree discrimination model is difficult to ensure by constructing a small number of labeled samples, and the balance between the training cost and the model precision is difficult to obtain. The method for identifying cheating behaviors based on the cheating rules set manually usually needs to introduce experts and priori knowledge in related fields to comb the cheating rules, the cheating rules are low in flexibility and not easy to expand, and for complex and changeable online scenes, the coverage rate of the cheating rules is usually remarkably reduced after a period of time.

In summary, how to identify cheating behaviors with low cost, high applicability and high accuracy has become a problem to be solved urgently at present.

Disclosure of Invention

In view of this, embodiments of the present application provide a user behavior data processing method, apparatus, device, and storage medium, which can implement low-cost, high-applicability, and high-accuracy identification of cheating behaviors.

A first aspect of the present application provides a method for processing user behavior data, where the method includes:

acquiring historical mode sequence data and current mode sequence data of a target user; the historical pattern sequence data is obtained by processing based on the historical behavior sequence data of the target user; the current mode sequence data is obtained by processing the current behavior sequence data of the target user;

performing feature extraction processing on the historical pattern sequence data through a target feature extraction model to obtain historical pattern features; performing feature extraction processing on the current mode sequence data through the target feature extraction model to obtain current mode features;

and determining whether the current behavior corresponding to the current behavior sequence data of the target user is a cheating behavior according to the similarity between the historical pattern feature and the current pattern feature.

A second aspect of the present application provides a user behavior data processing apparatus, including:

the sequence data acquisition module is used for acquiring historical mode sequence data and current mode sequence data of a target user; the historical pattern sequence data is processed based on historical behavior sequence data mining of the target user; the current mode sequence data is obtained by processing the current behavior sequence data of the target user;

the characteristic extraction module is used for carrying out characteristic extraction processing on the historical pattern sequence data through a target characteristic extraction model to obtain historical pattern characteristics; performing feature extraction processing on the current mode sequence data through the target feature extraction model to obtain current mode features;

and the cheating identification module is used for determining whether the current behavior corresponding to the current behavior sequence data of the target user is a cheating behavior according to the similarity between the historical pattern feature and the current pattern feature.

A third aspect of the present application provides an electronic device comprising a processor and a memory:

the memory is used for storing a computer program;

the processor is configured to execute the steps of the user behavior data processing method according to the first aspect.

A fourth aspect of the present application provides a computer-readable storage medium for storing a computer program for executing the steps of the user behavior data processing method according to the first aspect.

A fifth aspect of the present application provides a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to make the computer device execute the steps of the user behavior data processing method according to the first aspect.

According to the technical scheme, the embodiment of the application has the following advantages:

the embodiment of the application provides a user behavior data processing method, which comprises the steps of firstly acquiring historical mode sequence data and current mode sequence data of a target user, wherein the historical mode sequence data is obtained by processing the historical behavior sequence data of the target user, and the current mode sequence data is obtained by processing the current behavior sequence data of the target user; then, respectively carrying out feature extraction processing on the historical mode sequence data and the current mode sequence data through a target feature extraction model to obtain corresponding historical mode features and current mode features; and further, according to the similarity between the historical mode feature and the current mode feature, determining whether the current behavior corresponding to the current behavior sequence data of the target user is a cheating behavior. The historical pattern sequence data is obtained by processing the historical behavior sequence data of the target user and can reflect the operation habits of the target user to a certain extent, so that whether the current behavior of the target user accords with the past operation habits or not can be measured according to the similarity between the historical pattern features corresponding to the historical pattern sequence data and the current pattern features corresponding to the current pattern sequence data, and whether the current behavior of the user is cheating or not can be judged according to the similarity. Compared with the method for discovering the cheating behaviors based on the decision tree discrimination model, the method provided by the embodiment of the application does not need to acquire a large number of labeled samples to train the model special for identifying the cheating behaviors, so that the cost for identifying the cheating behaviors can be effectively reduced. Compared with a method for identifying cheating behaviors based on cheating rules set manually, the method provided by the embodiment of the application can effectively and accurately identify the cheating behaviors, and the scene applicability is high.

Drawings

Fig. 1 is a schematic view of an application scenario of a user behavior data processing method according to an embodiment of the present application;

fig. 2 is a schematic flowchart of a user behavior data processing method according to an embodiment of the present application;

FIG. 3 is a schematic structural diagram of a model to be trained according to an embodiment of the present disclosure;

fig. 4 is a schematic structural diagram of a first user behavior data processing apparatus according to an embodiment of the present application;

fig. 5 is a schematic structural diagram of a second user behavior data processing apparatus according to an embodiment of the present application;

fig. 6 is a schematic structural diagram of a third user behavior data processing apparatus according to an embodiment of the present application;

fig. 7 is a schematic structural diagram of a fourth user behavior data processing apparatus according to an embodiment of the present application;

fig. 8 is a schematic structural diagram of a fifth user behavior data processing apparatus according to an embodiment of the present application;

fig. 9 is a schematic structural diagram of a server according to an embodiment of the present application;

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

Detailed Description

In order to make the technical solutions of the present application better understood, 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 terms "first," "second," "third," "fourth," and the like in the description and in the claims of the present application and in the drawings described above, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

AI is a theory, method, technique and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technique of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a manner similar to human intelligence. Artificial intelligence is the research of the design principle and the realization method of various intelligent machines, so that the machines have the functions of perception, reasoning and decision making.

The artificial intelligence technology is a comprehensive subject and relates to the field of extensive technology, namely the technology of a hardware level and the technology of a software level. The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.

With the research and progress of artificial intelligence technology, the artificial intelligence technology is developed and applied in a plurality of fields, for example, common smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, unmanned driving, automatic driving, unmanned aerial vehicles, robots, smart medical treatment, smart customer service, and the like.

The common cheating behavior identification method in the related technology mainly comprises the following steps: a method for discovering cheating behaviors based on a decision tree discrimination model and a method for identifying the cheating behaviors based on cheating rules set manually. The method for discovering cheating behaviors based on the decision tree discrimination model needs to train the decision tree discrimination model special for identifying the cheating behaviors by using the labeled samples, the cost for constructing a large number of labeled samples is high, the model precision of the decision tree discrimination model is difficult to ensure by constructing a small number of standard samples, and therefore balance between the training cost and the model precision is difficult to obtain. The method for identifying cheating behaviors based on the cheating rules set manually needs to introduce experts and priori knowledge in related fields to comb the cheating rules, and the cheating rules are low in flexibility, not easy to expand and low in applicability to scenes.

In view of the problems in the related art, the embodiment of the present application provides a user behavior data processing method, which can be applied to various scenes in which cheating behaviors need to be identified, and has the advantages of high applicability, high accuracy, and no need of high cost.

Specifically, in the user behavior data processing method provided in the embodiment of the present application, historical pattern sequence data and current pattern sequence data of a target user are obtained first, where the historical pattern sequence data is obtained by processing the historical behavior sequence data of the target user, and the current pattern sequence data is obtained by processing the current behavior sequence data of the target user; then, respectively carrying out feature extraction processing on the historical mode sequence data and the current mode sequence data through a target feature extraction model to obtain corresponding historical mode features and current mode features; and further, according to the similarity between the historical mode feature and the current mode feature, determining whether the current behavior corresponding to the current behavior sequence data of the target user is a cheating behavior.

According to the user behavior data processing method, the operation habits of the target user are reflected by the historical mode sequence data obtained by processing the historical behavior sequence data based on the target user, and further, whether the current behavior of the target user accords with the past operation habits or not is measured according to the similarity between the historical mode characteristics corresponding to the historical mode sequence data and the current mode characteristics corresponding to the current mode sequence data, and whether the current behavior of the user is the cheating behavior or not is judged according to the similarity. Compared with the method for discovering the cheating behaviors based on the decision tree discrimination model, the method does not need to acquire a large number of labeled samples to train the model special for identifying the cheating behaviors, so that the cost for identifying the cheating behaviors can be effectively reduced. Compared with a method for identifying cheating behaviors based on cheating rules set manually, the method can effectively and accurately identify the cheating behaviors and is high in scene applicability.

It should be understood that the user behavior data processing method provided by the embodiment of the present application may be applied to an electronic device with data processing capability, such as a terminal device or a server. The terminal device may be a computer, a smart phone, a tablet computer, a Personal Digital Assistant (PDA), or the like; the server may specifically be an application server or a Web server, and in actual deployment, the server may be an independent server, or may also be a cluster server or a cloud server.

In order to facilitate understanding of the user behavior data processing method provided in the embodiment of the present application, an application scenario to which the user behavior data processing method is applicable is exemplarily described below by taking an execution subject of the user behavior data processing method as an example.

Referring to fig. 1, fig. 1 is a schematic view of an application scenario of a user behavior data processing method provided in an embodiment of the present application. As shown in fig. 1, the application scenario includes a terminal device 110 and a server 120, and communication between the terminal device 110 and the server 120 can be performed through a network. An application program runs in the terminal device 110, and a target user can execute a current behavior in a specific scene through the application program; for example, assuming that an online chess competition program runs in the terminal device 110, the target user may perform chess competition through the online chess competition program, and the behavior of the target user in the current competition process may be regarded as the current behavior of the target user. The server 120 is configured to execute the user behavior data processing method provided in the embodiment of the present application, and identify whether the target user is a cheating behavior according to the current operation executed by the terminal device 110.

In specific implementation, the target user may execute the current operation through the application program running in the terminal device 110, and after the terminal device 110 detects that the target user has executed the current operation, the terminal device may transmit operation data corresponding to the current operation to the server 120 through the network. After the preset time period is over, the server 120 may concatenate operation data corresponding to each operation performed by the target user within the preset time period to form current behavior sequence data of the target user, and convert the current behavior sequence data into corresponding current mode sequence data according to a preset policy identifier and/or a preset behavior identifier.

Still taking an application program running in the terminal device 110 as an online chess competition program as an example, after the target user finishes the next step of chess through the online chess competition program, the terminal device 110 may transmit operation data corresponding to the step of chess to the server 120, and after the target user finishes the field of chess competition, the server 120 may serially connect the operation data corresponding to each received step of operation to obtain current behavior sequence data of the target user in the field of chess competition, and further, according to a preset chess strategy identifier and/or a chess behavior identifier, convert the current progress as the sequence data into corresponding current mode sequence data.

Furthermore, the server 120 needs to acquire the historical pattern sequence data of the target user, which is mined based on the historical behavior sequence data of the target user, where the historical behavior sequence data corresponds to the historical behavior of the target user executed by the application program in the terminal device 110. Still taking an application program running in the terminal device 110 as an online chess competition program as an example, the historical behavior sequence data of the target user corresponds to a series of operation data of the target user in the historical chess competition, and the server 120 may perform data mining based on a plurality of pieces of historical behavior sequence data of the target user to obtain historical pattern sequence data of the target user, where the historical pattern sequence data can reflect the operation habits of the target user in the chess competition.

After acquiring the historical pattern sequence data and the current pattern sequence data of the target user, the server 120 may input the historical pattern sequence data into a pre-trained target feature extraction model to obtain the historical pattern feature corresponding to the historical pattern sequence data, and input the current pattern sequence data into the pre-trained target feature model to obtain the current pattern feature corresponding to the current pattern sequence data.

Furthermore, the server 120 may calculate a similarity between the historical pattern feature and the current pattern feature, where the similarity may reflect a matching degree between the current behavior of the target user and the past operation habits thereof, and if the similarity exceeds a preset similarity threshold, it may indicate that the current behavior of the target user conforms to the past operation habits thereof, and the current behavior is likely to be executed by the target user, so that the current behavior of the target user may be determined as not a cheating behavior.

It should be understood that the user behavior data processing method provided by the embodiment of the application can be applied to various scenes in which cheating behaviors need to be identified, such as identifying whether other users exist in a game to play the game instead of a target user, identifying whether players cheat in an online chess competition, identifying auxiliary attacks in an intelligence test, and the like. The application scenario of the user behavior data processing and identifying method provided by the embodiment of the present application is not limited at all.

Optionally, in another embodiment of the present application, in a sports game in which multiple users shoot online, if the user a cheats by using a plug-in software, user behavior data generated by the plug-in software is greatly different from a historical operating habit of the user a, and can be easily detected by the method provided in the embodiment of the present application, and various corresponding measures such as freezing or sealing a number or warning on an account number that is used for plug-in can be restricted according to the detection result.

The following describes the user behavior data processing method provided by the present application in detail through a method embodiment.

Referring to fig. 2, fig. 2 is a schematic flow chart of a user behavior data processing method according to an embodiment of the present application, and the following embodiment uses a server as an execution subject. As shown in fig. 2, the user behavior data processing method includes the following steps:

step 201: acquiring historical mode sequence data and current mode sequence data of a target user; the historical pattern sequence data is obtained by processing based on the historical behavior sequence data of the target user; the current mode sequence data is obtained by processing the current behavior sequence data of the target user.

When it is necessary to identify whether the current behavior of the target user is a cheating behavior, the server needs to acquire the history pattern sequence data and the current pattern sequence data of the target user. The historical pattern sequence data is processed based on the historical behavior sequence data of the target user, the historical behavior sequence data of the target user corresponds to the historical behavior operation of the target user, and the historical pattern sequence data processed based on the historical behavior sequence data of the target user can reflect the historical operation habit of the target user. The current mode sequence data is obtained by processing current behavior sequence data of the target user, and the current behavior sequence data of the target user corresponds to the current behavior operation of the target user.

Taking the case of applying the method provided by the embodiment of the present application to a scene of identifying whether a player cheats in an online chess game as an example, when a server needs to identify whether a player a cheats in a certain chess game, the server may obtain historical pattern sequence data and current pattern sequence data of the player a, the historical pattern sequence data of the player a being mined based on a plurality of pieces of historical behavior sequence data of the player a, the historical behavior sequence data of the player a corresponding to a past behavior operation of the player a in the chess game of the kind, the current pattern sequence data of the player a being obtained by converting the current behavior sequence data of the player a, the current behavior sequence data of the player a corresponding to the behavior operation of the player a in the chess game of the kind.

The following describes a manner of mining history pattern sequence data based on history behavior sequence data.

In specific implementation, the server may first obtain N (N is an integer greater than 1) pieces of historical behavior sequence data of the target user; then, based on a preset strategy identifier and/or a behavior operation identifier, converting each piece of acquired historical behavior sequence data into a corresponding historical behavior coding sequence; further, a target prefix set is constructed based on the N historical behavior coding sequences, the target prefix set comprises a plurality of target prefixes supporting a preset support degree condition, wherein the target prefix with the length of i is determined based on a projection data set corresponding to the target prefix with the length of i-1, and i is an integer larger than 1; finally, the target prefix with the longest length in the target prefix set can be determined as the historical pattern sequence data.

For example, when the server mines the historical pattern sequence data based on the historical behavior sequence data, all the historical behavior sequence data of the target user can be acquired. For example, if the server needs to identify whether the target user has cheating in a certain game, the server may obtain historical behavior sequence data corresponding to respective games in which the target user participates before the certain game, where the historical behavior sequence data corresponding to each game is obtained by concatenating each operation data of the target user in the certain game in chronological order.

Of course, in practical applications, the server may also obtain only part of the historical behavior sequence data of the target user, for example, obtain historical behavior sequence data generated by the target user in the last year or the last three months, and the number of the obtained historical behavior sequence data and the time corresponding to the historical behavior sequence data are not limited in any way in this application.

Then, the server may perform transcoding processing on each piece of acquired historical behavior sequence data based on a preset policy identifier and/or a behavior operation identifier to obtain a historical behavior coding sequence corresponding to each piece of historical behavior sequence data. For example, still taking the scenario of applying the method provided by the embodiment of the present application to identifying cheating of players in an online chess competition as an example, the server may preset respective identifiers corresponding to various strategies in the chess competition, for example, the identifier of the opening strategy 1 is set to be a1, the identifier of the opening strategy 2 is set to be a2, … …, the identifier of the defense strategy 1 is set to be B1, the identifier of the defense strategy 2 is set to be B2, and the like, and may also preset identifiers corresponding to various individual behavior operations in the chess competition, for example, the identifier corresponding to the behavior operation 1 is set to be a, the identifier corresponding to the behavior operation 2 is set to be B, and the like; then, for the sub-sequence data (which may correspond to a series of behavior operations or an independent behavior operation) in the historical behavior sequence data, converting the sub-sequence data (when the sub-sequence data corresponds to a series of behavior operations) into a corresponding policy identifier, or converting the sub-sequence data (when the sub-sequence data corresponds to an independent behavior operation) into a corresponding behavior operation identifier, and further combining the policy identifiers or the behavior identifiers corresponding to the sub-sequence data according to the arrangement order of the sub-sequence data in the historical behavior sequence data to obtain a historical behavior coding sequence corresponding to the historical behavior sequence data.

Furthermore, the server can adopt a prefixspan algorithm to mine implicit historical pattern sequence data based on the historical behavior code sequences corresponding to the N pieces of historical behavior sequence data. Specifically, the server may mine the historical pattern sequence data based on a minimum support degree policy, where a calculation method of the minimum support degree is shown in formula (1):

min_sup＝a×n (1)

wherein min _ sup is the minimum support degree; n is the number of the historical behavior sequence data, and N is equal to the N; a is a minimum support rate, which can be adjusted according to the number of the historical behavior sequence data and the number of the preset strategies.

When a target prefix set is specifically constructed, a server can find out behavior sequence elements with the length of 1 in N historical behavior coding sequences, and determine projection data sets corresponding to the found behavior sequence elements; then, counting the occurrence frequency of each behavior sequence element in the N historical behavior coding sequences, adding the behavior sequence elements with the occurrence frequency higher than the minimum support degree as target prefixes into a target prefix set, and continuously determining the target prefixes with the length of 2 based on the projection data sets corresponding to the target prefixes with the length of 1.

For a target prefix with a length i (where i is an integer greater than 1) and satisfying the minimum support requirement, the method can mine the following steps: and aiming at each target prefix with the length of i-1, determining the support degree corresponding to each projection data item in the corresponding projection data set, determining the projection data item with the support degree larger than the minimum support degree threshold value as the target projection data item, and combining the target prefix with the length of i-1 and the target projection data item to obtain the target prefix with the length of i.

That is, for a target prefix with a length i and satisfying the requirement of minimum support degree, the mining method can recursively mine the following ways: mining a projection data set corresponding to a target prefix with a certain length of i-1, and returning to recursion if the projection data set obtained by mining is an empty set; if the projection data set obtained by mining is not an empty set, counting the support degree (namely the occurrence frequency) of each projection data item in the corresponding projection data set, combining the projection data item meeting the requirement of the minimum support degree with the target prefix with the length of i-1 to obtain a target prefix with the length of i, and if the projection data item meeting the requirement of the minimum support degree does not exist, returning to recursion; and after finishing the operation aiming at each target prefix with the length of i-1, setting i to be i +1, and executing the operation again. In this way, all the target prefixes having a length of 1 to a length of i obtained in the above manner constitute a target prefix set.

The following describes a process of constructing a target prefix set corresponding to N historical behavior coding sequences, taking as an example that the N historical behavior coding sequences include a bcafgh and a bcdaghf, and the minimum support rate is 0.5.

For the historical behavior coding sequences bcafgh and bcdaghf, the target prefixes with the length of 1 which meet the minimum support requirement and the projection data sets corresponding to the target prefixes with the lengths of 1 are shown in table 1:

TABLE 1

Based on the projection data sets corresponding to the target prefixes with the length of 1, the server may further determine the target prefixes with the length of 2 that meet the minimum support requirement, and the projection data sets corresponding to the target prefixes with the length of 2 are shown in table 2:

TABLE 2

Based on the projection data sets corresponding to the target prefixes with the length of 2, the server may further determine the target prefixes with the length of 3 that meet the minimum support requirement, and the projection data sets corresponding to the target prefixes with the length of 3 are shown in table 3:

TABLE 3

Based on the projection data sets corresponding to the target prefixes with the length of 3, the server may further determine target prefixes with the length of 4 that meet the minimum support requirement, and the projection data sets corresponding to the target prefixes with the length of 4 are shown in table 4:

TABLE 4

Based on the projection data sets corresponding to the target prefixes with the length of 4, the server may further determine the target prefixes with the length of 5 that meet the minimum support requirement, and the projection data sets corresponding to the target prefixes with the length of 5 are shown in table 5:

TABLE 5

Length 4 target prefix	Corresponding projection data set
		bcagh	f

Each target prefix in tables 1 to 5 may construct a target prefix set, and since the target prefix with the longest length in the target prefix set can best reflect a series of behavior operations used by the target user, the server may finally select the target prefix with the longest length in the target prefix set as the historical pattern sequence data of the target user.

The history pattern sequence data may be temporarily mined by the server in response to the cheating recognition request, or may be previously mined by the server. For example, the server may acquire historical behavior sequence data of a target user in a past match when determining that the behavior of the target user in a certain match needs to be identified as a cheating behavior, and perform the above operation based on the acquired historical behavior sequence data to determine historical pattern sequence data of the target user; alternatively, the server may determine its corresponding history pattern sequence data for the target user in advance, and when the server needs to identify whether the target user's behavior in a certain game is a cheating behavior, the server may directly call the history pattern sequence data, and the server may update the history pattern sequence data periodically, for example, once a week or once a day, and the like. The present application does not limit the timing of determination of the above-described historical pattern sequence data at all.

The following describes a manner of converting the current pattern sequence data based on the current behavior sequence data of the target user.

In practical application, the method for converting the current behavior sequence data of the target user by the server is similar to the operation for converting the historical behavior sequence data into the historical behavior sequence data by the server, that is, the current behavior sequence data of the target user is subjected to code conversion processing based on the preset policy identifier and/or behavior operation identifier to obtain the current behavior sequence data corresponding to the current behavior sequence data, and the current behavior sequence data is the current mode sequence data. For details, reference may be made to the above related implementation manner for converting the historical behavior sequence data to obtain the corresponding historical behavior coding sequence, and details are not described here.

Optionally, in order to more accurately identify whether the current behavior of the target user is a cheating behavior, the server may obtain historical portrait feature data and current portrait feature data of the target user, in addition to the historical pattern sequence data and the current pattern sequence data of the target user, where the historical portrait feature data is determined according to the historical personal feature information and/or the historical behavior feature information of the target user, and the current portrait feature data is determined according to the current personal feature information and/or the current behavior feature information of the target user.

For example, taking a scene that the method provided by the embodiment of the present application is applied to identifying cheating of players in an online chess competition as an example, the server may construct historical portrait feature data of a target user according to historical personal feature information of the target user, such as age, gender, appearance, location, constellation and the like, and the server may also construct historical portrait feature data of the target user according to historical behavior information of the target user, such as average score, level of segment, chess playing character (such as conservation, robustness, violence and the like), habit number (such as abandoning and killing, city keeping, dead counterattack and the like), average duration of historical competition, and historical competition style (such as office type, aggressive type and the like).

Correspondingly, when the server constructs the current portrait feature data of the target user, the server can construct the current portrait feature data of the target user according to the feature information type according to which the historical portrait feature data is constructed and based on the current feature information of the target user under the feature information type.

Before the server specifically constructs the historical portrait feature data, it is usually required to determine which feature information is based on to construct the historical portrait feature data, at this time, for the candidate feature information, the server may determine a missing amount of the candidate feature information according to a configuration condition of each user in the target user group for the candidate feature information, and when the missing amount of the candidate feature information is less than a missing amount threshold, the candidate feature information may be determined as the feature information according to which the historical portrait feature data is generated, and the missing amount threshold is determined according to the number of users in the target user group and a preset missing rate.

For example, still taking the scenario that the method provided by the embodiment of the present application is applied to identify whether a player cheats in an online chess competition as an example, for some candidate feature information, the server may first determine the configuration of each user registered to use the online chess competition program for the candidate feature information, and take the number of users not configured with the candidate feature information as the missing amount of the candidate feature information. The server may calculate the miss amount threshold based on equation (2):

missing amount threshold is sample data amount × p (2)

The sample data size is the number of users in the target user group; p is a preset deficiency rate, and in practical application, p may be set according to a requirement of an actual scene, for example, may be set to 0.4.

If the missing amount of some candidate feature information is higher than the missing amount threshold, it indicates that too many users are not provided with the candidate feature information, and the candidate feature information is not suitable as a basis for generating the historical image feature data. If the missing amount of some candidate feature information is lower than the missing amount threshold, it is indicated that many users have the candidate feature information, and the candidate feature information can be used as the feature information for generating the historical image feature data.

Optionally, in order to ensure accuracy and reasonableness of the constructed historical image feature data, when the method provided by the embodiment of the application is used for constructing the historical image feature data for the target user, an abnormal value may be determined for the target feature information of the target user, and when the target feature information of the target user is determined to be an abnormal value, corresponding processing measures are taken to adjust the abnormal value.

Specifically, when the server determines the historical image feature data for the target user, it may determine whether the target feature information of the target user is an abnormal value, where the target feature information is feature information on which the historical image feature data is generated, and the abnormal value is determined according to a distribution of the target feature information of each user in the target user group. If the target feature information of the target user is determined to be an abnormal value, the server may determine, as the target feature information of the target user, a mean value of the target feature information of each user in the target user group in a case where the target feature information corresponds to the continuous type feature, and may determine, as the target feature information of the target user, a preset constant value in a case where the target feature information corresponds to the discrete type feature.

For example, for some target feature information, the server may sort the target feature information of each user in the target user group in a descending order or an ascending order, and then determine that the target feature information ranked at the first 1/m and/or the target feature information ranked at the last 1/m is an abnormal value, where the value of m may be set according to the actual scene requirement, and may be set to 10000, for example. If the server judges that the target characteristic information of the target user belongs to the abnormal value, the server can correct the target characteristic information of the target user; for example, when the target feature information corresponds to a continuous feature, the server may calculate a mean value of the target feature information of each user in the target user group, and then take the mean value as the target feature information of the target user; for another example, when such target feature information corresponds to a discrete type feature, the server may use a preset constant value as the target feature information of the target user.

It should be understood that, in practical applications, the server may determine the abnormal value of the target feature information in other manners besides the above manner, and the manner of determining the abnormal value of the feature information is not limited in this application. In addition, the server may adopt other ways to correct the abnormal target feature information besides the above way, according to the actual requirement, and the application does not make any limitation on the way of correcting the target feature information.

When the server generates historical portrait feature data of the target user according to the historical personal feature information and/or the historical behavior feature information of the target user, feature derivation processing can be performed on the basis of the existing historical personal feature information and/or historical behavior feature information, namely, feature derivation is performed through processing such as feature transformation, feature square, feature addition and subtraction and the like. For the continuous characteristic information, the server can perform box discretization, for the discrete characteristic information, the server can perform one-hot (one-hot) encoding processing on the discrete characteristic information, where one-hot encoding can also be referred to as one-bit effective encoding, which mainly adopts an N-bit state register to encode N states, each state corresponds to an independent register bit, and only one bit is effective at any time. Correspondingly, when the server generates the current portrait feature data of the target user according to the current personal feature information and/or the current behavior feature information of the target user, the generation mode is similar to the mode of generating the historical portrait feature data. The present application does not limit at all the way in which the server constructs the historical portrait characteristics data and the current portrait characteristics of the target user.

Step 202: performing feature extraction processing on the historical pattern sequence data through a target feature extraction model to obtain historical pattern features; and performing feature extraction processing on the current mode sequence data through the target feature extraction model to obtain current mode features.

After acquiring the historical pattern sequence data and the current pattern sequence data, the server may input the historical pattern sequence data into a pre-trained target feature extraction model to extract the historical pattern feature from the historical pattern sequence data through the target feature extraction model, and input the current pattern sequence data into the pre-trained target feature extraction model to extract the current pattern feature from the current pattern sequence data through the target feature extraction model.

It should be noted that, if the server obtains not only the historical pattern sequence data and the current pattern sequence data of the target user, but also the historical image feature data and the current image feature data of the target user in step 201; the server needs to perform fusion and splicing processing on the historical pattern sequence data and the historical portrait feature data to obtain historical fusion feature data, and then processes the historical fusion feature data through the target feature extraction model to obtain corresponding historical pattern features, and also needs to perform fusion and splicing processing on the current pattern sequence data and the current portrait feature data to obtain current fusion feature data, and then processes the current fusion feature data through the target feature extraction model to obtain corresponding current pattern features.

The following describes a training method of the target feature extraction model.

The server can obtain training sample data from a strategy knowledge base, wherein the strategy knowledge base is used for storing behavior sequence data and marking strategies with corresponding relations. And then, determining a prediction strategy result according to the behavior sequence data in the training sample data through a model to be trained, wherein the model to be trained consists of a feature extraction model to be trained and a classification model to be trained. And then, training the model to be trained according to the marking strategy in the training sample data and the prediction strategy result determined by the model to be trained. After the model to be trained satisfies the training end condition, the server may use the feature extraction model to be trained in the model to be trained as the target feature extraction model.

Taking the scenario that the method provided by the embodiment of the present application is applied to identifying whether players cheat in an online chess competition as an example, an exemplary description is given to the training mode of the target feature extraction model.

Specifically, the server can firstly carry out standardized and unified identification on game behaviors in the online chess competition, and different chess competitions have respective involving road numbers or methods, for example, in the chess competition, the opening strategy has Italy opening, double-horse defense, Hungary defense, Spanish opening, Western defense and the like, and the tactics have flash, double-general, catch-double, hold-back, lead-away and the like; the server may obtain a large number of chess game manuals, label certain strategies corresponding to the behavior sequence data in the chess game, and add the behavior sequence data labeled with the strategies to a strategy knowledge base to construct a strategy knowledge base, an exemplary strategy knowledge base is shown in table 6:

TABLE 6

Policy	Sequence of behaviors
		Strategy 1	Sequence of behaviors a
Strategy 1	Sequence of behaviors b
		Strategy 1	Sequence of behaviors c
Strategy 2	Sequence of behaviors d
		Strategy 2	Sequence of behaviors e
Strategy 2	Sequence of behaviors f
		Strategy 3	Sequence of behaviors g
Strategy 3	Sequence of behaviors h
		Strategy 3	Sequence of behaviors i
……	……

Furthermore, the server can use the data in the strategy knowledge base as training sample data to train the model to be trained. Illustratively, a multiple classification model (i.e., a model to be trained) based on margin loss can be constructed according to a form of a strategy knowledge base, and a Bi-LSTM (Bi-directional Long Short-Term Memory) model is introduced as a feature extraction model to be trained. Of course, in practical application, the Bi-LSTM model may be used as the feature extraction model to be trained, and other deep learning models such as Convolutional Neural Networks (CNNs), grus (gate recovery units), and the like may also be used as the feature extraction model to be trained, and the type of the feature extraction model to be trained is not limited in any way herein. The model structure of the model to be trained, which is constructed based on the Bi-LSTM model and the classification model to be trained am-softmax, is shown in FIG. 3, as shown in FIG. 3, the Bi-LSTM model comprises a plurality of parallel first sub-LSTM models 310, a plurality of parallel second sub-LSTM models 320 and a connection concat layer 330, wherein each first sub-LSTM model 310 is used for extracting the characteristics of input data from a first dimension, each second sub-LSTM model 320 is used for extracting the characteristics of the input data from a second dimension, and the concat layer 330 is used for splicing and fusing the characteristics extracted from the first sub-LSTM models and the characteristics extracted from the second sub-LSTM models; furthermore, the features fused by the concat layers 330 are processed by the full-link Fc layer 340, and are classified by am-softmax350 according to the features processed by the Fc layer 340.

The server trains the model to be trained based on a plurality of behavior sequences corresponding to each strategy in the strategy knowledge base, so that the model to be trained can identify a plurality of behavior sequences corresponding to the same strategy to belong to the same type. The server specifically trains the am-softmax-based multi-classification model, and can train based on the formula (3) and the formula (4):

y＝Bi-LSTM(x) (3)

p＝am-softmax(yW) (4)

wherein x is input behavior sequence data, and y is output of the Bi-LSTM model of the feature extraction model to be trained, namely features corresponding to the behavior sequence data. W is the corresponding policy label set, W ═ c₁，c₂，……，c_n)，c₁，c₂，……，c_nRespectively corresponding to various strategies in the strategy knowledge base, i.e. p ═ am-softmax (< y, c)_l＞，＜y，c₂＞，...，＜y，c_n＞。

The loss function for am-softmax is shown in equation (5):

wherein θ i represents y and c_iS and m are constants, s can be 30, and m can be 0.35.

When the model to be trained is trained, the model Bi-LSTM model to be trained and the classification model am-softmax to be trained in the model to be trained can be trained by minimizing the loss function of the am-softmax. After determining that the model to be trained satisfies the preset training end condition, for example, determining that the iterative training frequency of the model to be trained reaches a preset training frequency threshold, or determining that the model accuracy of the model to be trained reaches a preset accuracy threshold, the feature extraction model to be trained in the model to be trained may be used as the target feature extraction model in the present application.

Optionally, in the process of applying the target feature extraction model, the policy knowledge base may be expanded by using the behavior sequence data processed by the target feature extraction model. Specifically, the server may calculate a similarity between the behavior sequence data processed by the feature extraction model and the existing behavior sequence data in the policy knowledge base, and for the behavior sequence data with the similarity exceeding a preset similarity threshold, the server may label the policy corresponding to the existing behavior sequence data, and further store the labeled behavior sequence data in the policy knowledge base to expand the policy knowledge base.

Step 203: and determining whether the current behavior corresponding to the current behavior sequence data of the target user is a cheating behavior according to the similarity between the historical pattern feature and the current pattern feature.

After the server performs the above operations, the server performs feature extraction processing on the historical pattern sequence data and the current pattern sequence data respectively by using the target feature extraction model to obtain corresponding historical pattern features and current pattern features, and then the similarity between the current pattern features and the historical pattern features can be further calculated. And then, judging whether the similarity exceeds a preset similarity threshold, if so, indicating that the current behavior of the target user accords with the historical operation habit of the target user, and therefore, determining that the current behavior of the target user is not a cheating behavior, and if not, indicating that the current behavior of the target user does not accord with the historical operation habit of the target user, and therefore, determining that the current behavior of the target user belongs to the cheating behavior.

In the cheating behavior recognition method provided by the embodiment of the application, the operation habits of the target user are reflected by the historical pattern sequence data obtained by mining the historical behavior sequence data based on the target user, and then whether the current behavior of the target user accords with the past operation habits or not is measured according to the similarity between the historical pattern features corresponding to the historical pattern sequence data and the current pattern features corresponding to the current pattern sequence data, so that whether the current behavior of the user is a cheating behavior or not is judged according to the similarity. Compared with the method for discovering the cheating behaviors based on the decision tree discrimination model, the method does not need to acquire a large number of labeled samples to train the model special for identifying the cheating behaviors, so that the cost for identifying the cheating behaviors can be effectively reduced. Compared with a method for identifying cheating behaviors based on cheating rules set manually, the method can effectively and accurately identify the cheating behaviors for a long time, and is high in scene applicability.

For the user behavior data processing method described above, the present application also provides a corresponding user behavior data processing apparatus, so that the user behavior data processing method is applied and implemented in practice.

Referring to fig. 4, fig. 4 is a schematic structural diagram of a user behavior data processing apparatus 400 corresponding to the user behavior data processing method shown in fig. 2, where the user behavior data processing apparatus 400 includes:

a sequence data acquisition module 401, configured to acquire historical pattern sequence data and current pattern sequence data of a target user; the historical pattern sequence data is processed based on historical behavior sequence data mining of the target user; the current mode sequence data is obtained by processing the current behavior sequence data of the target user;

a feature extraction module 402, configured to perform feature extraction processing on the historical pattern sequence data through a target feature extraction model to obtain a historical pattern feature; performing feature extraction processing on the current mode sequence data through the target feature extraction model to obtain current mode features;

a cheating identifying module 403, configured to determine whether a current behavior corresponding to the current behavior sequence data of the target user is a cheating behavior according to a similarity between the historical pattern feature and the current pattern feature.

Optionally, on the basis of the user behavior data processing apparatus shown in fig. 4, referring to fig. 5, fig. 5 is a schematic structural diagram of another user behavior data processing apparatus 500 provided in the embodiment of the present application. As shown in fig. 5, the apparatus further includes:

a data mining module 501, configured to obtain N pieces of historical behavior sequence data of the target user, where N is an integer greater than 1; converting each piece of historical behavior sequence data into a corresponding historical behavior coding sequence based on a preset strategy identifier and/or behavior operation identifier; constructing a target prefix set based on the N historical behavior coding sequences; the target prefix set comprises a plurality of target prefixes meeting a preset support degree condition, wherein the target prefix with the length of i is determined based on a projection data set corresponding to the target prefix with the length of i-1, and i is an integer greater than 1; and determining the target prefix with the longest length in the target prefix set as the historical pattern sequence data.

Optionally, on the basis of the user behavior data processing apparatus shown in fig. 5, the data mining module 501 specifically determines the target prefix with the length i by the following means:

aiming at each target prefix with the length of i-1, determining the support degree corresponding to each projection data item in the corresponding projection data set, and determining the projection data item with the corresponding support degree larger than the minimum support degree threshold value as a target projection data item; and combining the target prefix with the length of i-1 and the target projection data single item to obtain the target prefix with the length of i.

Optionally, on the basis of the user behavior data processing apparatus shown in fig. 4, referring to fig. 6, fig. 6 is a schematic structural diagram of another user behavior data processing apparatus 600 provided in the embodiment of the present application. As shown in fig. 6, the apparatus further includes:

the model training module 601 is used for acquiring training sample data from the strategy knowledge base; the strategy knowledge base is used for storing behavior sequence data and marking strategies with corresponding relations; determining a prediction strategy result according to the behavior sequence data in the training sample data through a model to be trained; the model to be trained consists of a feature extraction model to be trained and a classification model to be trained; training the model to be trained according to the labeling strategy and the prediction strategy result in the training sample data; and after the model to be trained is determined to meet the preset training end condition, taking the feature extraction model to be trained in the model to be trained as the target feature extraction model.

Optionally, on the basis of the user behavior data processing apparatus shown in fig. 4, referring to fig. 7, fig. 7 is a schematic structural diagram of another user behavior data processing apparatus 700 provided in the embodiment of the present application. As shown in fig. 7, the apparatus further includes:

a portrait data acquisition module 701, configured to acquire historical portrait feature data and current portrait feature data of the target user; the historical portrait characteristic data is determined according to historical personal characteristic information and/or historical behavior characteristic information of the target user; the current portrait characteristic data is determined according to current personal characteristic information and/or current behavior characteristic information of the target user;

the feature extraction module 402 is specifically configured to:

performing fusion splicing processing on the historical pattern sequence data and the historical portrait feature data to obtain historical fusion feature data; performing feature extraction processing on the historical fusion feature data through the target feature extraction model to obtain the historical mode features;

performing fusion splicing processing on the current mode sequence data and the current portrait feature data to obtain current fusion feature data; and performing feature extraction processing on the current fusion feature data through the target feature extraction model to obtain the current mode feature.

Optionally, on the basis of the user behavior data processing apparatus shown in fig. 7, referring to fig. 8, fig. 8 is a schematic structural diagram of another user behavior data processing apparatus 800 provided in the embodiment of the present application. As shown in fig. 8, the apparatus further includes:

an image feature information selecting module 801, configured to determine, for candidate feature information, a missing amount of the candidate feature information according to a configuration situation of each user in a target user group for the candidate feature information; when the missing amount of the candidate feature information is smaller than a missing amount threshold value, determining the candidate feature information as feature information according to which the historical portrait feature data are generated; the missing amount threshold is determined according to the number of users in the target user group and a preset missing rate.

Optionally, on the basis of the user behavior data processing apparatus shown in fig. 7, the portrait data obtaining module 701 determines the historical portrait feature data specifically by:

when the historical portrait feature data is determined for the target user, judging whether target feature information of the target user is an abnormal value; the target characteristic information is characteristic information which is used for generating the historical portrait characteristic data; the abnormal value is determined according to the distribution condition of the target characteristic information of each user in the target user group;

if so, determining the average value of the target feature information of each user in the target user group as the target feature information of the target user under the condition that the target feature information corresponds to the continuous feature; determining a preset constant value as the target feature information of the target user in a case where the target feature information corresponds to a discrete type feature.

The user behavior data processing device provided in the embodiment of the application reflects the operation habit of the target user by using the historical pattern sequence data obtained by mining the historical behavior sequence data based on the target user, and further measures whether the current behavior of the target user conforms to the past operation habit or not according to the similarity between the historical pattern feature corresponding to the historical pattern sequence data and the current pattern feature corresponding to the current pattern sequence data, so as to determine whether the current behavior of the user is a cheating behavior or not.

The embodiment of the present application further provides an electronic device for identifying cheating behaviors, where the electronic device may be specifically a server or a terminal device, and the server and the terminal device provided in the embodiment of the present application will be described below from the perspective of hardware materialization.

Referring to fig. 9, fig. 9 is a schematic structural diagram of a server 900 according to an embodiment of the present disclosure. The server 900 may vary widely in configuration or performance and may include one or more Central Processing Units (CPUs) 922 (e.g., one or more processors) and memory 932, one or more storage media 930 (e.g., one or more mass storage devices) storing applications 942 or data 944. Memory 932 and storage media 930 can be, among other things, transient storage or persistent storage. The program stored on the storage medium 930 may include one or more modules (not shown), each of which may include a series of instruction operations for the server. Still further, a central processor 922 may be provided in communication with the storage medium 930 to execute a series of instruction operations in the storage medium 930 on the server 900.

The Server 900 may also include one or more power supplies 926, one or more wired or wireless network interfaces 950, one or more input-output interfaces 958, and/or one or more operating systems, such as a Windows Server^TM，Mac OS X^TM，Unix^TM,Linux^TM，FreeBSD^TMAnd so on.

The steps performed by the server in the above embodiments may be based on the server structure shown in fig. 9.

The CPU 922 is configured to execute the following steps:

Optionally, the CPU 922 may also be configured to execute steps of any implementation manner of the user behavior data processing method provided in the embodiment of the present application.

Referring to fig. 10, fig. 10 is a schematic structural diagram of a terminal device according to an embodiment of the present application. For convenience of explanation, only the parts related to the embodiments of the present application are shown, and details of the specific technology are not disclosed. Taking a terminal as a smart phone as an example:

fig. 10 is a block diagram illustrating a partial structure of a smart phone related to a terminal provided in an embodiment of the present application. Referring to fig. 10, the smart phone includes: radio Frequency (RF) circuit 1010, memory 1020, input unit 1030, display unit 1040, sensor 1050, audio circuit 1060, wireless fidelity (WiFi) module 1070, processor 1080, and power source 1090. Those skilled in the art will appreciate that the smartphone configuration shown in fig. 10 is not intended to be limiting and may include more or fewer components than shown, or some components in combination, or a different arrangement of components.

The memory 1020 may be used to store software programs and modules, and the processor 1080 executes various functional applications and data processing of the smart phone by operating the software programs and modules stored in the memory 1020. The memory 1020 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data, a phonebook, etc.) created according to the use of the smartphone, and the like. Further, the memory 1020 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.

The processor 1080 is a control center of the smartphone, connects various parts of the entire smartphone through various interfaces and lines, and executes various functions and processes data of the smartphone by running or executing software programs and/or modules stored in the memory 1020 and calling data stored in the memory 1020, thereby integrally monitoring the smartphone. Optionally, processor 1080 may include one or more processing units; preferably, the processor 1080 may integrate an application processor, which handles primarily the operating system, user interfaces, applications, etc., and a modem processor, which handles primarily the wireless communications. It is to be appreciated that the modem processor described above may not be integrated into processor 1080.

In the embodiment of the present application, the processor 1080 included in the terminal further has the following functions:

Optionally, the processor 1080 is further configured to execute the steps of any implementation manner of the user behavior data processing method provided in the embodiment of the present application.

The embodiment of the present application further provides a computer-readable storage medium, configured to store a computer program, where the computer program is configured to execute any one implementation manner of the user behavior data processing method described in the foregoing embodiments.

Embodiments of the present application also provide a computer program product or computer program comprising computer instructions stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes any one of the implementation manners of the user behavior data processing method described in the foregoing embodiments.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing computer programs.

It should be understood that in the present application, "at least one" means one or more, "a plurality" means two or more. "and/or" for describing an association relationship of associated objects, indicating that there may be three relationships, e.g., "a and/or B" may indicate: only A, only B and both A and B are present, wherein A and B may be singular or plural. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. "at least one of the following" or similar expressions refer to any combination of these items, including any combination of single item(s) or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", wherein a, b, c may be single or plural.

The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.

Claims

1. A method for processing user behavior data, the method comprising:

2. The method of claim 1, wherein the historical pattern sequence data is obtained by:

acquiring N pieces of historical behavior sequence data of the target user, wherein N is an integer greater than 1;

converting each piece of historical behavior sequence data into a corresponding historical behavior coding sequence based on a preset strategy identifier and/or behavior operation identifier;

constructing a target prefix set based on the N historical behavior coding sequences; the target prefix set comprises a plurality of target prefixes meeting a preset support degree condition, wherein the target prefix with the length of i is determined based on a projection data set corresponding to the target prefix with the length of i-1, and i is an integer greater than 1;

and determining the target prefix with the longest length in the target prefix set as the historical pattern sequence data.

3. The method of claim 2, wherein the target prefix of length i is determined by:

4. The method of claim 1, wherein the target feature extraction model is trained by:

acquiring training sample data from a strategy knowledge base; the strategy knowledge base is used for storing behavior sequence data and marking strategies with corresponding relations;

determining a prediction strategy result according to the behavior sequence data in the training sample data through a model to be trained; the model to be trained consists of a feature extraction model to be trained and a classification model to be trained;

training the model to be trained according to the labeling strategy and the prediction strategy result in the training sample data;

and after the model to be trained is determined to meet the preset training end condition, taking the feature extraction model to be trained in the model to be trained as the target feature extraction model.

5. The method of claim 1, further comprising:

acquiring historical portrait feature data and current portrait feature data of the target user; the historical portrait characteristic data is determined according to historical personal characteristic information and/or historical behavior characteristic information of the target user; the current portrait characteristic data is determined according to current personal characteristic information and/or current behavior characteristic information of the target user;

performing feature extraction processing on the historical pattern sequence data through a target feature extraction model to obtain historical pattern features, wherein the feature extraction processing includes:

performing fusion splicing processing on the historical pattern sequence data and the historical portrait feature data to obtain historical fusion feature data;

performing feature extraction processing on the historical fusion feature data through the target feature extraction model to obtain the historical mode features;

performing feature extraction processing on the current mode sequence data through the target feature extraction model to obtain current mode features, including:

performing fusion splicing processing on the current mode sequence data and the current portrait feature data to obtain current fusion feature data;

and performing feature extraction processing on the current fusion feature data through the target feature extraction model to obtain the current mode feature.

6. The method of claim 5, further comprising:

for candidate feature information, determining the missing amount of the candidate feature information according to the configuration condition of each user in a target user group aiming at the candidate feature information;

when the missing amount of the candidate feature information is smaller than a missing amount threshold value, determining the candidate feature information as feature information according to which the historical portrait feature data are generated; the missing amount threshold is determined according to the number of users in the target user group and a preset missing rate.

7. The method of claim 5, wherein the historical representation feature data is determined by:

8. A user behavior data processing apparatus, characterized in that the apparatus comprises:

the sequence data acquisition module is used for acquiring historical mode sequence data and current mode sequence data of a target user; the historical pattern sequence data is obtained by processing based on the historical behavior sequence data of the target user; the current mode sequence data is obtained by processing the current behavior sequence data of the target user;

9. An electronic device, comprising a processor and a memory;

the memory is used for storing a computer program;

the processor is configured to execute the user behavior data processing method according to any one of claims 1 to 7 in accordance with the computer program.

10. A computer-readable storage medium for storing a computer program for executing the user behavior data processing method according to any one of claims 1 to 7.