Data processing method and system

Abstract

The embodiment of the invention provides a data processing method and system. In the method, after historical behavior data corresponding to a plurality of known users with preset behavior patterns are obtained, M initial behavior sequences are generated based on the time sequence of each behavior in a plurality of behaviors of one user in the historical behavior data. The population evolution process is simulated for the M initial behavior sequences, multi-generation breeding is conducted on the initial population, N target behavior sequences are obtained, and the characterization capacity of the N target behavior sequences to a preset behavior model is higher than that of the M initial behavior sequences. By means of the mode, mining of the initial behavior sequence is automatically determined, so that the target behavior sequence with the high characterization capacity for the preset behavior model is determined.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a data processing method and system. Background Art

[0002] With the development of artificial intelligence technology, more and more fields have the need to provide personalized services for different groups of people. By analyzing the user's behavior data, users can be divided into different groups according to the needs of the actual scenario, so as to provide different services for different groups of people.

[0003] Usually, in order to achieve population segmentation, it is necessary to predefine the identification features corresponding to different populations. For example, for a population known to have a certain behavior pattern, the behavior data of users in the population is manually analyzed and the rules are summarized to obtain the common behavior sequences of users in the population, and these behavior sequences are used as the identification features corresponding to the behavior pattern. Then, the identification features are used to identify the population with the behavior pattern from a large number of users.

[0004] However, the above method requires manual analysis and summary of a large number of user behavior data, which is inefficient, especially when dealing with large-scale data sets. In addition, manual summary methods can usually only obtain some simple explicit features of the crowd, and the expressive power of these characteristic behavior patterns is limited, resulting in inaccurate results of crowd segmentation, which in turn affects service quality.

[0005] The content of the background technology section is only the information known to the inventor personally, and does not mean that the above information has entered the public domain before the filing date of this specification, nor does it mean that it can become the prior art of this specification. Summary of the invention

[0006] The present specification provides a data processing method and system, which can automatically mine the identification features corresponding to the preset behavior pattern based on the historical behavior data of a target user group with a preset behavior pattern by simulating population evolution.

[0007] In a first aspect, the present specification provides a data processing method, comprising: obtaining historical behavior data corresponding to a target user group, the target user group comprising a plurality of users known to have preset behavior patterns; generating M initial behavior sequences based on the historical behavior data, wherein each initial behavior sequence represents a plurality of behaviors performed in chronological order by a user in the target user group, and M is an integer greater than 1; using the M initial behavior sequences as an initial population, breeding the initial population for multiple generations by simulating a population evolution process to obtain a target population, the target population comprising N target behavior sequences, the N target behavior sequences having a higher ability to represent the preset behavior pattern than the M initial behavior sequences; and determining, based on each target behavior sequence in the target population, an identification feature corresponding to the preset behavior pattern.

[0008] In a second aspect, the present specification also provides a data processing system, comprising at least one storage medium and at least one processor, wherein the at least one storage medium stores at least one instruction set for performing a data processing method; the at least one processor is communicatively connected to the at least one storage medium, wherein the at least one processor reads the at least one instruction set when running, and executes according to the instructions of the at least one instruction set: the data processing method described in any one of the above-mentioned first aspect.

[0009] In a third aspect, the present specification also provides a computer-readable non-volatile storage medium, wherein the computer-readable non-volatile storage medium stores at least one instruction set, and when the at least one instruction set is executed by at least one processor, it implements the data processing method as described in any one of the first aspects above.

[0010] It can be seen from the above technical solutions that the data processing method and system provided in this specification, after obtaining the historical behavior data of the target user group known to have a preset behavior pattern, generates M initial behavior sequences based on the historical behavior data. The M initial behavior sequences are used as the initial population, and the initial population is reproduced for multiple generations by simulating the population evolution process to obtain the target population. The characterization ability of the behavior sequence in the target population for the preset behavior pattern is higher than the characterization ability of the initial behavior sequence for the preset behavior pattern, ensuring that the target population can accurately characterize the preset behavior pattern. And the multi-generation reproduction method can generate a variety of behavior sequences to ensure that the behavioral characteristics of different users can be captured. Through population evolution, the implicit characteristics and complex relationships between the characteristics of multiple dimensions can be discovered. Based on each target behavior sequence in the target population, the identification features of the preset behavior pattern are determined. Through the identification features, the user group with the preset behavior pattern can be matched more accurately, and the accuracy of the subsequent user group division can be improved. And the above data processing methods are all automated, which improves the efficiency of data processing and reduces the subjectivity and deviation of human judgment.

[0011] Other functions of the data processing method and system provided in this specification will be partially listed in the following description. The creative aspects of the data processing method and system provided in this specification can be fully explained by practicing or using the methods, devices and combinations described in the following detailed examples. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of this specification, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0013] Figure 1 A schematic diagram of an application scenario provided according to an embodiment of this specification is shown;

[0014] Figure 2 A schematic diagram of the hardware structure of a computing device provided according to some embodiments of this specification is shown;

[0015] Figure 3 A schematic diagram of a data processing method according to an embodiment of the present specification is shown;

[0016] Figure 4 A schematic diagram of a breeding process according to another embodiment of the present specification is shown;

[0017] Figure 5 A schematic diagram of a multi-generation breeding process according to an embodiment of this specification is shown;

[0018] Figure 6 A schematic diagram of a single-point crossover algorithm provided according to an embodiment of this specification is shown;

[0019] Figure 7 A schematic diagram showing a multi-point crossover algorithm provided according to an embodiment of the present specification; and

[0020] Figure 8 A schematic diagram of a mutation algorithm provided according to an embodiment of the present specification is shown. DETAILED DESCRIPTION

[0021] The following description provides specific application scenarios and requirements of this specification, with the purpose of enabling those skilled in the art to make and use the contents of this specification. Various local modifications to the disclosed embodiments will be apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of this specification. Therefore, this specification is not limited to the embodiments shown, but to the widest scope consistent with the claims.

[0022] The terms used herein are only used for the purpose of describing specific example embodiments and are not restrictive. For example, unless the context clearly indicates otherwise, as used herein, the singular forms "a", "an" and "the" may also include plural forms. When used in this specification, the terms "include", "comprise" and / or "contain" mean that the associated integers, steps, operations, elements and / or components exist, but do not exclude the existence of one or more other features, integers, steps, operations, elements, components and / or groups or that other features, integers, steps, operations, elements, components and / or groups may be added in the system / method.

[0023] In view of the following description, these and other features of the present specification, as well as the operation and function of the related elements of the structure, and the economy of the combination and manufacture of the parts can be significantly improved. Reference is made to the accompanying drawings, all of which form a part of this specification. However, it should be clearly understood that the drawings are for illustration and description purposes only and are not intended to limit the scope of this specification. It should also be understood that the drawings are not drawn to scale.

[0024] The flowcharts used in this specification illustrate the operations implemented by the system according to some embodiments in this specification. It should be clearly understood that the operations of the flowcharts may not be implemented in sequence. On the contrary, the operations may be implemented in reverse order or simultaneously. In addition, one or more other operations may be added to the flowchart. One or more operations may be removed from the flowchart.

[0025] In this specification, "X includes at least one of A, B or C" means that X includes at least A, or X includes at least B, or X includes at least C. That is, X may include only any one of A, B, and C, or may include any combination of A, B, and C and other possible contents / elements at the same time. The any combination of A, B, and C may be A, B, C, AB, AC, BC, or ABC.

[0026] In this specification, unless explicitly stated otherwise, the association relationship between structures can be a direct association relationship or an indirect association relationship. For example, when describing "A is connected to B", unless it is explicitly stated that A is directly connected to B, it should be understood that A can be directly connected to B or indirectly connected to B; for another example, when describing "A is above B", unless it is explicitly stated that A is directly above B (AB is adjacent and A is above B), it should be understood that A can be directly above B or indirectly above B (AB is separated by other elements and A is above B). And so on.

[0027] It should be noted that the user data obtained in this manual is authorized by the user and does not involve user privacy.

[0028] It should be noted that the collection, storage, use, processing, transmission, provision and disclosure of relevant information of users involved in the technical solution of this specification are in compliance with the relevant laws and regulations and do not violate public order and good morals.

[0029] For the convenience of description, the terms that will appear in the following text of this specification are first explained.

[0030] Genetic Algorithm (GA) is a search and optimization algorithm inspired by natural selection and genetic mechanisms. It simulates operations such as selection, crossover (recombination) and mutation in the biological evolution process to find the optimal solution or approximate optimal solution to the problem.

[0031] User behavior refers to the various activities and interaction patterns that users exhibit in a specific environment or system. These behaviors can include various operations and interactions of users in digital environments such as websites, applications, social media platforms, and e-commerce platforms.

[0032] Behavior sequence refers to a series of behaviors performed by users in a specific environment or system in chronological order. These behaviors may include clicks, browsing, purchases, searches, sharing, etc. By recording the chronological order of these behaviors, an orderly behavior sequence can be formed.

[0033] Tracking: It is a data collection technology that records user behavior data at specific locations by inserting code (i.e. "tracking") at key locations in applications or websites. This data can include user clicks, swipes, inputs, page dwell time, and other behaviors, as well as basic user information (such as user ID, device information, etc.).

[0034] Crossover operation: In genetic algorithms, crossover is an important operation used to generate new offspring individuals. The crossover operation generates new offspring individuals by combining the genes of two parent individuals.

[0035] The technical solution provided in this specification can be applied to scenarios where it is necessary to divide the population. For example, in a personalized recommendation scenario, by dividing the population, different content can be recommended based on the preferences of different groups of people. For another example, in a risk control scenario, by dividing the population, high-risk groups can be accurately located, thereby achieving key prevention and control of high-risk groups.

[0036] Taking the personalized recommendation scenario as an example, in order to achieve population segmentation, it is necessary to predefine the identification features corresponding to different populations. For example, for a population known to have a certain behavior pattern, the behavior data of users in the population is manually analyzed and the rules are summarized to obtain the common behavior sequences of users in the population, and these behavior sequences are used as the identification features corresponding to the behavior pattern. Then, the identification features are used to identify the population with the behavior pattern from a large number of users. This method of manually analyzing and summarizing the behavior sequence not only consumes a lot of manpower and time costs, but also has a low processing speed. It is impossible to achieve a thousand faces for a thousand people, and ignores the logical relationship between user behavior and results. In addition, the manual processing method can only find strongly correlated sequences, and it is difficult to mine weakly correlated but effective behavior sequences. It is impossible to deeply understand the needs of users, which may lead to inaccurate analysis results and affect the quality of subsequent decision-making.

[0037] The technical solution provided in this specification, after obtaining the historical behavior data corresponding to the target user group, can generate M initial behavior sequences based on the historical behavior data, and use the M initial behavior sequences as the initial population. Among them, each behavior sequence represents multiple behaviors performed by a user in chronological order. This method of determining the initial population based on historical behavior data ensures the diversity of the initial population. Subsequently, the technical solution provided in this specification reproduces the initial population for multiple generations by simulating the population evolution process, retains the target behavior sequence with better performance, removes some behavior sequences with poor performance, and obtains a target population with a higher representation ability for the preset behavior pattern than the initial population. Finally, the technical solution provided in this specification will also determine the identification features corresponding to the preset behavior pattern based on each target behavior sequence in the target population. This method of determining the identification features based on the target population, because the target population has a strong representation ability for the preset behavior ability, the obtained identification features are also more representative of the preset behavior pattern. In other words, the technical solution provided in this specification can not only improve the efficiency and accuracy of determining the identification features, but also the obtained identification features can better reflect the user's behavior pattern.

[0038] It should be noted that the above description of the application scenario is only one of the multiple usage scenarios provided in this specification. Those skilled in the art should understand that when the data processing method and system provided in this specification are applied to other usage scenarios, their implementation methods and technical effects are similar.

[0039] Figure 1 A schematic diagram of an application scenario 100 provided according to an embodiment of this specification is shown.

[0040] like Figure 1As shown, the application scenario 100 may include a feature mining system 130 and a user mining system 150. The application scenario 100 involves two execution stages. Among them, in the first stage, the feature mining system 130 can mine the identification features corresponding to the preset behavior pattern by analyzing the historical behavior data of multiple users (target user groups) known to have a preset behavior pattern. Among them, the preset behavior pattern refers to the behavior pattern or behavior sequence that occurs repeatedly by an individual or group in a certain situation. For example, the preset behavior pattern may include but is not limited to: buying a cup of coffee after buying breakfast, exercising at lunch time, accustomed to buying high-value items, accustomed to buying discounted items, frequent large transfers, etc. In the second stage, the user mining system 150 can mine the user group with the preset behavior pattern from the user set according to the above-mentioned identification features, thereby realizing the selection (mining) of users with the preset behavior pattern based on the identification features. It can be understood that the above-mentioned target user group can be a user group composed of a small number of users known to have a preset behavior pattern, and the user group mined by the user mining system 150 from the user set can be a user group composed of more users with a preset behavior pattern.

[0041] The feature mining system 130 may be a computing system with certain computing capabilities. The feature mining system 130 may execute the data processing method provided in this specification, thereby extracting identification features corresponding to preset behavior patterns from the historical behavior data corresponding to the target user group. The feature mining system 130 may store data or instructions for executing the data processing method described in this specification, and may execute or be used to execute data or instructions. The feature mining system 130 may include a hardware device with data information processing functions and the necessary programs required to drive the hardware device to work.

[0042] The feature mining system 130 may be a single computing device or a cluster system composed of multiple computing devices. The data or instructions stored in the feature mining system 130 for executing the data processing method described in this specification may adopt any form of system architecture, such as a layered architecture, an event-driven architecture, a micro-core architecture, a micro-service architecture, or a cloud architecture.

[0043] In some embodiments, the feature mining system 130 may first generate M initial behavior sequences based on the historical behavior data corresponding to the target user group, and use the M initial behavior sequences as the initial population. The feature mining system 130 then reproduces the initial population for multiple generations by simulating the population evolution process to obtain the target population. The target population includes N target behavior sequences, and the characterization capabilities of the N target behavior sequences for the preset behavior patterns are higher than the characterization capabilities of the M initial behavior sequences for the preset behavior patterns. Finally, the feature mining system 130 may determine the identification features corresponding to the preset behavior patterns based on each target behavior sequence in the target population.

[0044] That is to say, the feature mining system 130 can obtain the identification features corresponding to the preset behavior pattern based on the historical behavior data mining of the target user group. Compared with the method of manually combing the historical behavior data of the user and dividing the population, the technical solution provided in this specification does not need to be manually combed to perform user mining, and the user mining system 150 can be directly used to mine users with preset behavior patterns from the user set. Subsequently, the user mining system 150 automatically mines users with preset behavior patterns from the user set based on the identification features. Therefore, the technical solution provided in this specification improves the speed of determining the target population, thereby improving the speed of determining the identification features corresponding to the users with the preset behavior pattern, and reducing the consumption of manpower costs and time costs caused by manual combing. And in the embodiment of this specification, the user mining system 150 determines the target behavior sequence through multi-generation reproduction, which can not only find strongly related behavior sequences, but also find some weakly related, but better behavior sequences with the ability to characterize the preset behavior. The accuracy and diversity of the target behavior sequences contained in the target population are improved, thereby improving the accuracy of the identification features corresponding to the preset behavior patterns determined based on the target population in the future.

[0045] In addition, since the identification features mined by the feature mining system 130 are the identification features corresponding to the preset behavior pattern, they are identification features with high characterization capabilities for the preset behavior pattern and have directionality. This can avoid the influence of other identification features (identification features unrelated to the preset behavior pattern) on user mining, thereby making the users with the preset behavior pattern mined from the user set more accurate.

[0046] The user mining system 150 and the feature mining system 130 may correspond to the same computing system or may correspond to different computing systems respectively, which is not limited in this specification.

[0047] In some embodiments, the data processing system provided in this specification may correspond to Figure 1 In some embodiments, the data processing system provided in this specification may include Figure 1 The feature mining system 130 and the user mining system 150 in.

[0048] Figure 2 A hardware structure diagram of a computing device 200 provided according to some embodiments of this specification is shown. The computing device 200 can be used as Figure 1 In some embodiments, when the feature mining system 130 uses a device cluster, the computing device 200 can be used as any device in the feature mining system 130.

[0049] like Figure 2 As shown, computing device 200 includes at least one storage medium 230 and at least one processor 220. In some embodiments, computing device 200 may further include an internal communication bus 210. In some embodiments, computing device 200 may further include a communication port 250. In some embodiments, computing device 200 may further include an I / O component 260.

[0050] Internal communication bus 210 may connect various system components, including storage media 230 and processor 220. I / O components 260 support input / output between computing device 200 and other components.

[0051] The communication port 250 is used for data communication between the computing device 200 and the outside world. For example, the computing device 200 can be connected to a network through the communication port 250.

[0052] The storage medium 230 may include a data storage device. The data storage device may be a non-temporary storage medium or a temporary storage medium. For example, the data storage device may include one or more of a disk 232, a read-only storage medium (ROM) 234, or a random access storage medium (RAM) 236. The storage medium 230 also includes at least one instruction set stored in the data storage device. The instruction set is a computer program code, and the computer program code may include a program, a routine, an object, a component, a data structure, a process, a module, etc. for executing the data processing method provided in this specification.

[0053] At least one processor 220 is connected to at least one storage medium 230 through an internal communication bus 210. At least one processor 220 is used to execute the at least one instruction set. When the system 130 is running, at least one processor 220 reads at least one instruction set and executes the data processing method provided in this specification according to the instructions of at least one instruction set.

[0054] The processor 220 may perform all steps included in the data processing method. The processor 220 may be in the form of one or more processors. The processor 220 may issue execution instructions. The processor 220 may include one or more hardware processors, such as a microcontroller, a microprocessor, a reduced instruction set computer (RISC), an application-specific integrated circuit (ASIC), an application-specific instruction set processor (ASIP), a central processing unit (CPU), a graphics processing unit (GPU), a physical processing unit (PPU), a microcontroller unit, a digital signal processor (DSP), a field programmable gate array (FPGA), an advanced RISC machine (ARM), a programmable logic device (PLD), any circuit or processor capable of performing one or more functions, etc., or any combination thereof.

[0055] For illustrative purposes only, only one processor 220 is shown in the computing device 200 in the attached drawings in this specification. However, it should be noted that the computing device 200 in this specification may also include multiple processors, and therefore, the operations and / or method steps disclosed in this specification may be performed by one processor as in this specification, or may be performed jointly by multiple processors. For example, if the processor 220 of the computing device 200 in this specification performs step A and step B, it should be understood that step A and step B may also be performed jointly or separately by two different processors 220 (e.g., the first processor performs step A, the second processor performs step B, or the first and second processors perform steps A and B together).

[0056] Figure 3 FIG. 3 is a flow chart of a data processing method provided according to an embodiment of the present specification. The data processing method P300 may be executed by the feature mining system 130. Figure 3 As shown, the method P300 provided in this specification may include S310-S370, wherein:

[0057] S310: Obtaining historical behavior data corresponding to a target user group, where the target user group includes a plurality of users known to have preset behavior patterns.

[0058] In the embodiments of this specification, the preset behavior patterns may include: frequently engaging in high-risk behaviors, habitually purchasing discounted goods, habitually purchasing high-value goods, habitually drinking coffee every morning, etc. It should be understood that the above embodiments are only exemplary, and the specific preset behavior patterns may include contents that can be flexibly adjusted according to user needs, and are not limited to the above embodiments. The application scenarios may also be health management, online education, or content consumption, etc., which are determined according to actual application conditions.

[0059] When determining the historical behavior data corresponding to the target user group, the corresponding historical behavior data can be determined based on the current application scenario and the target user group. For example, the historical behavior data in an e-commerce scenario can be a user's behavior log. The historical behavior data in a social media scenario can be a user's interaction record. The historical behavior data in a financial platform scenario can be a user's transaction record, etc. Obtaining the historical behavior data of the target user group includes but is not limited to obtaining the user's identifier (ID), behavior type, behavior time, behavior object, and other information.

[0060] S330: Generate M initial behavior sequences based on the historical behavior data, where each initial behavior sequence represents a plurality of behaviors performed by a user in the target user group in chronological order, and M is an integer greater than 1.

[0061] In actual applications, the user's historical behavior data is usually collected based on the data collection points pre-deployed in the code. These data collection points can be called embedded points. For example, taking the social scenario as an example, a embedded point can be set in advance in the code segment used to respond to the collection operation in the social program. When the code segment is triggered to execute, it means that the user has performed the collection operation, and the collection operation is recorded as the user's behavior data. The data collected by each embedded point can correspond to a user's behavior. The data collected by different embedded points can be arranged in chronological order to obtain the user's behavior sequence.

[0062] In some possible embodiments, a large number of data collection points (buried points) may be widely deployed in an application scenario (in an application or website) to record various user behaviors. These buried points can cover almost all interactive behaviors of users during use. However, some of the massive buried points may not be representative enough of the key behavioral capabilities of users and are not highly correlated with the user's behavior patterns. Therefore, in order to reduce the impact of irrelevant data, some representative and valuable key buried points can be pre-determined in the massive buried points based on preset behavior patterns.

[0063] For example, in a shopping scenario, massive tracking points may include: page loading tracking points, login / registration tracking points, personal data modification tracking points, delivery address filling tracking points, page scrolling tracking points, search submission tracking points, product description reading tracking points, product click tracking points, add to shopping cart click tracking points, buy now click tracking points, navigation menu click tracking points, promotion activity click tracking points, product sharing click tracking points, view similar products click tracking points, etc.

[0064] If the preset behavior pattern of the target user group is to purchase high-value goods on the e-commerce platform, then obviously the page loading points (highly correlated with the user's current network status), login / registration points, delivery address filling points, personal data modification points, etc. are not highly correlated with the preset behavior pattern, and there is no need to obtain the behavior data corresponding to the above points.

[0065] That is to say, in the above scenario, the key buried points (the number is L) selected from the massive buried points may include: page scroll buried points, search submission buried points, product description reading buried points, add to shopping cart click buried points, buy immediately click buried points, promotion activity click buried points, product sharing click buried points, view similar product click buried points, etc. It should be understood that the buried point types and key buried points that may be included in the above massive buried points can be flexibly adjusted according to the current application scenario, user needs and actual conditions, and are not limited to the above embodiments.

[0066] In order to ensure the validity of the data, before extracting the initial behavior sequence based on the historical behavior data, the historical behavior data can be preprocessed to remove some invalid data, redundant data and abnormal data in the historical behavior data to ensure that the extracted data is accurate and complete.

[0067] In some possible embodiments, a large amount of behavioral data may be generated when a user uses an application or website, which may include many irrelevant or repetitive behaviors. If the analysis is based on a large amount of behavioral data of each user in the target user group, not only the computational efficiency is low, the amount of computation is large, and there may be a risk of overfitting. Therefore, in order to improve the computational efficiency, the behavioral data of M users can be randomly obtained from the historical behavioral data of a large number of target user groups, and a key behavior set corresponding to the preset behavior pattern can be obtained from the behavioral data of the above M users (the key behavior set is determined based on the key buried points, one buried point corresponds to one behavior, and multiple key behaviors corresponding to multiple key buried points constitute a key behavior set). The key behavior set includes multiple key behaviors related to the preset behavior pattern. Then, for each of the M users, multiple behaviors of the user in the key behavior set are selected from the user's behavior data, and an initial behavior sequence is generated based on the multiple behaviors.

[0068] That is to say, the embodiment of this specification first randomly obtains the behavior data of M users from the historical behavior data of a large number of target user groups. Compared with directly obtaining the behavior data of all target user groups, the workload in the subsequent data processing process is reduced to a certain extent. Subsequently, the technical solution provided in this specification does not obtain the complete behavior data of each user, but screens the behavior data of each user. Under a large amount of behavior data of a user, only the key behavior set corresponding to the preset behavior pattern is obtained based on the key points, which reduces the workload in the subsequent data processing process to a certain extent.

[0069] In order to further improve the efficiency of subsequent data processing, in one embodiment of the present specification, the behavior data in all key behavior sets are not obtained. Instead, when determining a set of initial behavior sequences, multiple behaviors in the key behavior set are extracted by extracting multiple key behaviors from the key behavior set. And based on the time sequence of the multiple behaviors, the multiple behaviors are sorted to form an initial behavior sequence corresponding to the user. In other words, an initial behavior sequence may include P behaviors performed by a user in chronological order, and the P behaviors are all from the key behavior set, and P is an integer greater than 1 and less than L (the number of key buried points). This method of extracting multiple behaviors in the key behavior set not only reduces data noise, but also reduces the workload of subsequent data processing and improves computing efficiency.

[0070] In the embodiments of the present specification, in order to ensure the diversity of the initial population and to ensure that the behavior sequence can contain more different types of behaviors, the method of taking P key points in the key points can be random sampling. That is, based on the historical behavior data of each user in the target user group, P key behaviors are randomly selected from the key behavior set corresponding to the preset behavior pattern of each user, and each key behavior is sorted according to the execution time of each key behavior in the P key behaviors to obtain the initial behavior sequence of each user. The above method of extracting multiple key behaviors, since multiple key behaviors are extracted from the key behavior set, ensures the representativeness of the obtained behavior sequence. Since the extraction method of multiple key behaviors is random sampling, the diversity of the obtained behavior sequence is guaranteed. Thereby ensuring the quality of the obtained initial population.

[0071] Still taking the key buried points (number is L) as an example, it may include: page scroll buried points, search submission buried points, product description reading buried points, add to shopping cart click buried points, buy now click buried points, promotion activity click buried points, product sharing click buried points, and view similar product click buried points. For example, P may be 3, then the P buried points randomly selected each time may be any 3 of the above L key buried points. For example, the M initial behavior sequences may include initial behavior sequence 1, initial behavior sequence 2 and initial behavior sequence 3, where initial behavior sequence 1 is composed of the above three behavior data of searching for products, adding to shopping carts, and clicking "buy now". Initial behavior sequence 2 is composed of viewing product details, sharing products, and clicking "buy now". Initial behavior sequence 3 is composed of adding to shopping carts, clicking products, and sharing products. It should be understood that the above embodiments are only exemplary descriptions, and the specific number settings of L and P, as well as the actual situation of buried points, can be flexibly adjusted according to user needs, and are not limited to those given in the above embodiments.

[0072] In the embodiments of the present specification, after generating M initial behavior sequences based on historical behavior data, each initial behavior sequence can be real-coded to facilitate subsequent operations and processing, and facilitate subsequent operations such as crossover and mutation. Real-code can compress multi-dimensional discrete data into one-dimensional or low-dimensional continuous data, reduce the problem of dimensional explosion, reduce the complexity of subsequent calculations, and improve calculation efficiency.

[0073] In the embodiments of this specification, a corresponding real number code is set in advance for each behavior to ensure that the real number code of each behavior is unique in the behavior sequence, which is convenient for subsequent calculation and analysis. A corresponding real number code is set for each behavior by pre-creating a mapping table. For example, each behavior and its corresponding real number code in a shopping scenario are shown in Table 1 below.

[0074] Table 1

[0075] Behavior Real number encoding Search for products 1 Click on the product 2 View product details 3 Share products 4 add to the cart 5 Visit the shopping cart page 6 Click "Buy Now" 7 Fill in the delivery address 8 Confirm payment 9 Payment successful 10

[0076] Among them, the encoding method is to convert each behavior data in the initial behavior sequence into a real number according to a preset mapping relationship, so as to convert each initial behavior sequence into a real number sequence. Still taking initial behavior sequence 1, initial behavior sequence 2 and initial behavior sequence 3 as examples for explanation, the real number sequence corresponding to initial behavior sequence 1 [search for products, add to shopping cart, click "buy now"] is [1, 5, 7]. The real number sequence corresponding to initial behavior sequence 2 [view product details, share products, click "buy now"] is [3, 4, 7,]. The real number sequence corresponding to initial behavior sequence 3 [add to shopping cart, click on products, share products] is [5, 2, 4]. Through the above method, the string sequence is converted into a real number sequence, the data format in the behavior sequence is unified, the efficiency of data processing is improved, and the complexity and inefficiency of string operations are avoided. It should be understood that the above embodiments are only exemplary and are not limited to the above embodiments.

[0077] S350: Taking the M initial behavior sequences as the initial population, multiplying the initial population for multiple generations by simulating the population evolution process to obtain the target population, wherein the target population includes N target behavior sequences, and the representation capability of the N target behavior sequences for the preset behavior patterns is higher than the representation capability of the M initial behavior sequences for the preset behavior patterns.

[0078] In the embodiments of this specification, in order to ensure the accuracy of the target population, multiple generations of reproduction may be used to continuously iterate until the target population is obtained.

[0079] Figure 4 A schematic diagram of a breeding process according to another embodiment of the present specification is shown. Figure 4 As shown, multi-generation reproduction may include a total of X generations of reproduction, each generation of reproduction corresponds to a population. During the reproduction process, the ability of the behavior sequence in the population to represent the preset behavior pattern is gradually improved until the target population is obtained. The target population has a higher ability to represent the preset behavior model than the initial population.

[0080] Figure 5 A schematic diagram of a multi-generation breeding process according to an embodiment of this specification is shown, such as Figure 5 As shown, the i-th generation reproduction process in multi-generation reproduction includes S351-S355, where:

[0081] S351: Selecting some behavior sequences that meet the evolution condition from the current population to form a first behavior sequence set, where the evolution condition is a condition related to the ability to represent the preset behavior pattern.

[0082] Among them, the fitness of each behavior sequence can be calculated based on the fitness function. The fitness function uses the behavior sequence as an independent variable and the fitness as a dependent variable. The fitness obtained based on the fitness function characterizes the ability of a behavior sequence to characterize the preset behavior pattern. Different preset behavior patterns correspond to different fitness functions. For example, if the application scenario is a lending scenario, and the problem to be optimized is to increase the borrowing rate, then the fitness function should be positively correlated with the borrowing rate. If the current application scenario is a shopping scenario, and the problem to be optimized is to increase the purchase rate, then the fitness should be positively correlated with the purchase rate. That is to say, in the embodiments of the present specification, the fitness function corresponding to the preset behavior pattern is first obtained, and then based on the fitness function, the fitness corresponding to each behavior sequence in the current population is determined. Based on the fitness corresponding to each behavior sequence in the current population, some behavior sequences that meet the evolution conditions are selected from the current population to form a first behavior sequence set.

[0083] In some possible embodiments, the evolution condition may be a preset fitness threshold, and a method of selecting some behavior sequences from the current population based on fitness may be to select some behavior sequences whose fitness meets the preset fitness threshold from the current population.

[0084] In order to avoid the problem that the fitness values ​​between behavior sequences are too different, some behavior sequences with particularly high fitness quickly occupy most of the population, resulting in a decrease in genetic diversity and premature convergence to the local optimal solution. In the embodiments of this specification, the fitness function can be resized using an exponential transformation to standardize the fitness to a uniform interval, thereby reducing the problem of too large a fitness difference between different behavior sequences.

[0085] After setting the fitness function, each behavior sequence in the current population is evaluated by the fitness function to quantify the quality of user behavior in each group of behavior sequences. The higher the fitness, the better the ability of the behavior sequence to represent the preset behavior pattern. The lower the fitness, the worse the ability of the behavior sequence to represent the preset behavior pattern.

[0086] In the process of multi-generation reproduction, there may be some infeasible solutions. That is, although some behavior sequences meet the requirements of the preset fitness threshold, they violate the constraints of the actual situation. Therefore, such behavior sequences cannot be regarded as meeting the requirements of the preset fitness threshold. For example, behavior sequence 7 [share products, fill in the delivery address, search for products] may obtain a very high fitness value and meet the fitness threshold, but when the preset behavior mode is to purchase high-value goods, it is obvious that the above behavior sequence 7 violates the actual situation. Few people buy high-value goods after performing the above behavior operations, which has no practical significance.

[0087] In order to solve the above problems, the feasibility of the behavior sequence is taken into consideration in the embodiments of the present specification. By introducing a penalty term, the penalty term is configured to adjust the fitness of a behavior sequence based on the feasibility of the behavior sequence, so that the fitness of an infeasible solution (such as the above-mentioned behavior sequence 7) is much lower than any feasible solution. When the feasibility of the first behavior sequence is lower than the feasibility of the second behavior sequence, the adjusted fitness corresponding to the first behavior sequence is lower than the adjusted fitness corresponding to the second behavior sequence. For example, the fitness value of the penalty term can be set to a very low preset fitness value. The above-mentioned method of introducing the penalty term greatly reduces the fitness value of the infeasible solution in the subsequent multi-generation reproduction process, prompting it to be gradually eliminated, or gradually transformed into a feasible solution.

[0088] In some possible embodiments, the evolution condition may also be a preset probability threshold, and the evolution process of each behavior sequence in the current population may be determined by a Roulette Wheel Selection method. The core idea of ​​this method is to determine the probability of each behavior sequence being selected into the next generation population according to its fitness. The higher the fitness of a behavior sequence, the greater the probability of being selected, and vice versa. Specifically, assuming that the number of behavior sequences included in the current population is K, then the i-th behavior sequence a i The probability of being selected is p i is the behavior sequence a i The fitness f i The sum of the fitness of all individuals in the current population That is to say,

[0089] When selecting a portion of the behavior sequences that meet the evolution conditions, for example, random selection is performed based on multiple behavior sequences that meet the evolution conditions to determine the portion of the behavior sequences that meet the evolution conditions. Alternatively, multiple behavior sequences may be sorted according to their probabilities of being selected, and the portion of the behavior sequences that are ranked top may be determined based on the sorting results. Among them, the number of partial behavior sequences may be, for example, a preset constant value. Or it may be a preset ratio, and the specific number is determined based on the number of behavior sequences included in the current population and the preset ratio. It should be understood that the above embodiments are only exemplary, and the specific method of determining the portion of the behavior sequences that meet the evolution conditions can be flexibly adjusted according to user needs, and is not limited to the above embodiments.

[0090] In the embodiment of this specification, the fitness mean corresponding to the current population can be determined based on the fitness corresponding to each behavior sequence in the current population. According to the fitness mean and each behavior sequence a in the current population i The fitness f i The ratio of the two determines the selection probability p of each behavior sequence.i That is to say, A behavior sequence with a selection probability greater than a preset probability threshold is selected from the current population to form a first behavior sequence set. It should be understood that the above embodiment is only an exemplary description, and the specific evolution conditions and how to determine the selection probability of each behavior sequence can be flexibly adjusted according to user needs and are not limited to the above embodiment.

[0091] S353: performing a crossover operation on different behavior sequences in the first behavior sequence set to obtain a plurality of evolved behavior sequences, and adding the plurality of evolved behavior sequences to the first behavior sequence set to obtain a second behavior sequence set.

[0092] Among them, since the fitness of the behavior sequences included in the first behavior sequence set is relatively high and meets the evolutionary conditions, it means that the behavior sequences included in the first behavior sequence set have a better characterization effect on the preset behavior pattern, and such excellent behavior sequences should inherit their characteristics to the next generation. Therefore, in the embodiment provided in this specification, the first behavior sequence set is directly copied to the second behavior sequence set, and a crossover operation is performed based on the first behavior sequence set to obtain multiple evolved behavior sequences based on the first behavior sequence set. In other words, the second behavior sequence set is composed of the first behavior sequence set and the multiple evolved behavior sequences obtained after crossover based on the first behavior sequence set.

[0093] In the embodiments of the present specification, the evolving behavior sequence includes multiple subsequences, wherein the multiple subsequences are respectively from at least two different behavior sequences in the first behavior sequence set. That is to say, the number of behavior sequences participating in a crossover operation is at least 2, and may be 3 or even more. The number of behavior sequences participating in a crossover operation can be flexibly adjusted according to user needs and is not limited to the above embodiments.

[0094] In the following embodiments, the number of behavior sequences participating in a crossover operation is 2 as an example: for example, 2 behavior sequences can be randomly determined in the first behavior sequence as parent behavior sequences, and part of the behavior data in the parent behavior sequence can be exchanged according to a preset crossover rate to obtain a child behavior sequence, and the obtained child behavior sequence is added to the second behavior sequence set.

[0095] According to the number of crossover points, the crossover operation can be divided into single-point crossover operation and multi-point crossover operation. Figure 6 FIG. 2 shows a schematic diagram of a single-point crossover algorithm provided according to an embodiment of the present specification. Figure 6 As shown, the single-point crossover operation can randomly determine a crossover point in the parent behavior sequence according to a preset crossover rate, and exchange part of the behavior data in the two parent behavior sequences at the crossover point to generate two child behavior sequences.

[0096] like Figure 6 As shown, an exemplary description of a single-point crossover operation can be: assuming there are two parent behavior sequences A and parent behavior sequence B, the behavior data included are: parent behavior sequence A (behavior 1, behavior 2, behavior 3, behavior 4, behavior 5, behavior 6, behavior 7, behavior 8); parent behavior sequence B (behavior 8, behavior 7, behavior 6, behavior 5, behavior 4, behavior 3, behavior 2, behavior 1). The position after the second behavior in the sequence is taken as the intersection point, and part of the behavior data before the intersection point (of course, it can also be after the intersection point) is exchanged. To generate two child behavior sequences: child behavior sequence 1 (behavior 8, behavior 7, behavior 3, behavior 4, behavior 5, behavior 6, behavior 7, behavior 8); child 2 behavior sequence (behavior 1, behavior 2, behavior 6, behavior 5, behavior 4, behavior 3, behavior 2, behavior 1).

[0097] Figure 7 FIG. 2 shows a schematic diagram of a multi-point crossover algorithm provided according to an embodiment of the present specification. Figure 7 As shown, the multi-point crossover operation can select multiple crossover points in the parent behavior sequence, and then exchange part of the behavior data of the two parent behavior sequences at the multiple crossover points to generate two new child behavior sequences.

[0098] Still taking the above two parent behavior sequences A and parent behavior sequence B as examples for illustrative description, in the multi-point crossover operation, for example, the position after the second behavior and the position after the fifth behavior in the sequence are used as intersections. After exchanging part of the behavior data outside the two intersections (of course, it can also be between the two intersections), two child behavior sequences are generated: child behavior sequence 3 (behavior 8, behavior 7, behavior 3, behavior 4, behavior 5, behavior 3, behavior 2, behavior 1), child behavior sequence 4 (behavior 1, behavior 2, behavior 6, behavior 5, behavior 4, behavior 6, behavior 7, behavior 8). It should be understood that the above embodiments are only illustrative descriptions. In the embodiments of this specification, only a single-point crossover operation or only a multi-point crossover operation may be used, or a combination of a single-point crossover operation and a multi-point crossover operation may be used. The determination of the specific crossover method can be flexibly adjusted according to the application scenario and user needs, and this specification does not impose any restrictions here.

[0099] S355: Generate a population after the i-th generation of reproduction based on the second behavior sequence set, where i is any natural number. When i=1, the current population is the initial population, and when i>1, the current population is the population after the i-1-th generation of reproduction.

[0100] In the embodiment of the present description, the method of generating the population after the i-th generation of reproduction can be: by performing mutation operation on some behavior sequences in the second behavior sequence set, a plurality of mutated behavior sequences are obtained. The plurality of mutated behavior sequences and other behavior sequences in the second behavior sequence set that have not been mutated are used as the population after the i-th generation of reproduction.

[0101] The mutation operation may be one or more of the following: randomly adding a preset number of behaviors to the behavior sequence; randomly deleting a preset number of behaviors in the behavior sequence; or randomly replacing some behaviors in the behavior sequence with other behaviors. The method for determining the preset number may be related to the mutation rate. The higher the mutation rate, the larger the value of the preset number, and the lower the mutation rate, the smaller the value of the preset number. Mutating some behavior sequences can create new behavior sequences and prevent the algorithm from shrinking to a local optimal solution. The specific mutation rate setting can be flexibly adjusted according to user needs and is not limited to the above embodiments.

[0102] For example, taking the parent behavior sequence C (behavior 1, behavior 2, behavior 6, behavior 5, behavior 3) as an example, for the mutation operation of randomly adding a preset number of behaviors, the mutation method may be: randomly selecting a position in the parent behavior sequence to add a new behavior. In one embodiment, a random behavior may be randomly added after the first behavior in the parent behavior sequence C through the mutation operation to obtain the mutated child behavior sequence C1 (behavior 1, behavior 5, behavior 2, behavior 6, behavior 5, behavior 3).

[0103] For the mutation operation of randomly deleting a preset number of behaviors in the behavior sequence, the mutation method may be: randomly deleting behaviors at 2 positions in the parent behavior sequence. In one embodiment, the first and third behaviors may be randomly deleted from the parent behavior sequence C through the mutation operation to obtain the mutated child behavior sequence C1 (behavior 2, behavior 5, behavior 3).

[0104] Figure 8 A schematic diagram of a mutation algorithm provided according to an embodiment of the present specification is shown. Figure 8 The mutation operation shown in is to randomly replace some behaviors in the behavior sequence with other behaviors. Figure 8As shown, still taking the parent behavior sequence C as an example for explanation, its variation mode may be to randomly replace the behaviors in the two parent behavior sequences with other random behaviors. In one embodiment, the first and fourth behaviors in the parent behavior sequence C may be randomly mutated, and the mutated child behavior sequence C3 (behavior 9, behavior 2, behavior 6, behavior 8, behavior 3). It should be understood that the above embodiment is only an exemplary description, and the specific preset number of settings, the position of the randomly determined mutated behavior, etc. can be adjusted according to the actual situation or user needs, and are not limited to the above embodiment.

[0105] In the embodiment of the present specification, the termination condition of multi-generation reproduction includes at least one of the following: the number of reproduction iterations is greater than or equal to the preset number of iterations T max That is, the current number of iterations T is greater than or equal to the preset number of iterations T max (T>T max ), then the multi-generation reproduction ends.

[0106] The termination condition of multi-generational reproduction may also be: the characterization ability of the behavior sequences contained in the current population for the preset behavior pattern reaches the convergence condition. The convergence condition can be characterized by introducing an evaluation function. When the change of the evaluation function is less than the preset threshold, it can be considered that the characterization ability of the current population has converged, and the multi-generational reproduction is terminated. It should be understood that the above embodiment is only an exemplary description, and the specific termination condition of multi-generational reproduction may also be that each behavior sequence in the current population reaches a preset fitness threshold, or the diversity of the current population drops to a preset threshold, etc. The specific termination condition of multi-generational reproduction can be flexibly adjusted according to user needs, and is not limited to the above embodiment.

[0107] S370: Determine the identification features corresponding to the preset behavior patterns based on the target behavior sequences in the target population.

[0108] After determining the identification features, users with behavior patterns can be subsequently mined from the user set based on the identification features. For example, for each user in the user set, the user's behavior features can be determined based on the user's historical behavior data, and the similarity between the user's behavior features and the above identification features can be determined. When the similarity is greater than or equal to a preset threshold, it is determined that the user has the above preset behavior pattern. When the similarity is less than the preset threshold, it is determined that the user does not have the above preset behavior pattern. In the above manner, users with preset behavior patterns can be mined from the user set.

[0109] In the personalized recommendation application scenario, more people with preset behavior patterns are mined from the user set to facilitate subsequent accurate recommendations to achieve personalized recommendations or targeted recommendations. In the market application scenario, user segmentation can be achieved through the above method, which is convenient for targeted market strategies. In other possible application scenarios, different user management strategies or user maintenance strategies can be adopted according to the behavior patterns of different users to improve user retention and repurchase rates. Or in the risk identification scenario, potential fraud can also be discovered and prevented in a timely manner by identifying the identification features of abnormal behavior patterns. It should be understood that after mining users with behavior patterns from the user set, how to perform subsequent processing can be flexibly adjusted according to the current application scenario and the actual needs of the user, and is not limited to the above embodiments.

[0110] In summary, the data processing method and system provided in this specification, after obtaining the historical behavior data of the target user group known to have a preset behavior pattern, generates M initial behavior sequences based on the historical behavior data. The M initial behavior sequences are used as the initial population, and the initial population is reproduced for multiple generations by simulating the population evolution process to obtain the target population. The characterization ability of the behavior sequence in the target population for the preset behavior pattern is higher than the characterization ability of the initial behavior sequence for the preset behavior pattern, ensuring that the target population can accurately characterize the preset behavior pattern. And the multi-generation reproduction method can generate a variety of behavior sequences to ensure that the behavioral characteristics of different users can be captured. Through population evolution, the implicit characteristics and complex relationships between features of multiple dimensions can be discovered. Based on each target behavior sequence in the target population, the identification features of the preset behavior pattern are determined. Through the identification features, the user group with the preset behavior pattern can be matched more accurately, and the accuracy of the subsequent user group division can be improved. And the above-mentioned data processing methods are all automated, which improves the efficiency of data processing and reduces the subjectivity and deviation of human judgment.

[0111] In addition, in the method provided in this specification, all historical behavior data are not processed, but only the behavior data of M users are randomly obtained from the historical behavior data, and the behavior data of M users are screened to obtain a key behavior set. Then some behaviors in the key behavior set are extracted to obtain M behavior sequences in the initial population. The processing efficiency in the subsequent data processing process is greatly improved and the performance consumption is reduced. In the breeding process, this specification reproduces the initial population for multiple generations through multiple breeding methods such as selection (determination of the first behavior sequence set), crossover (crossover operation to obtain the evolutionary behavior sequence) and mutation (mutated behavior sequence), so as to finally obtain a target behavior sequence with a good characterization effect on the preset behavior pattern. Finally, the recognition features corresponding to the preset behavior pattern are determined based on the target behavior sequence, which significantly improves the accuracy of the recognition features, the generalization ability of the model and the accuracy of user behavior prediction. At the same time, the data processing efficiency is improved and human errors are reduced.

[0112] On the other hand, this specification provides a computer-readable non-temporary storage medium, which stores at least one instruction set for data processing executable instructions. When at least one instruction set is executed by a processor, at least one instruction set instructs the processor to implement the steps of the data processing method P300 of this specification. In some possible implementations, various aspects of this specification can also be implemented in the form of a program product, which includes program code. When the program product is run on the system, the program code is used to make the data processing system perform the steps of the method P300 described in this specification. The program product for implementing the above method can use a portable compact disk read-only memory (CD-ROM) to include program code and can be run on a data processing system. However, the program product of this specification is not limited to this. In this specification, the readable storage medium can be any tangible medium containing or storing a program, which can be used by the instruction execution system or used in combination with it. The program product can use any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media include: an electrical connection with one or more conductors, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. A computer-readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above. The program code for performing the operations of this specification can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" language or similar programming languages.

[0113] The above is a description of a specific embodiment of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require a specific order or a continuous order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0114] In summary, after reading this detailed disclosure, those skilled in the art will appreciate that the foregoing detailed disclosure may be presented only by way of example and may not be limiting. Although not explicitly stated herein, those skilled in the art will appreciate that this specification requires various reasonable changes, improvements, and modifications to the embodiments. These changes, improvements, and modifications are intended to be proposed by this specification and are within the spirit and scope of the exemplary embodiments of this specification.

[0115] In addition, certain terms in this specification have been used to describe embodiments of this specification. For example, "one embodiment", "an embodiment" and / or "some embodiments" mean that a particular feature, structure or characteristic described in conjunction with the embodiment may be included in at least one embodiment of this specification. Therefore, it can be emphasized and should be understood that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various parts of this specification do not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics may be appropriately combined in one or more embodiments of this specification.

[0116] It should be understood that in the foregoing description of the embodiments of this specification, in order to help understand a feature and for the purpose of simplifying this specification, this specification combines various features in a single embodiment, figure or its description. However, this does not mean that the combination of these features is necessary. When reading this specification, it is entirely possible for a person skilled in the art to mark out some of the devices as separate embodiments. In other words, the embodiments in this specification can also be understood as the integration of multiple secondary embodiments. And the content of each secondary embodiment is also valid when it is less than all the features of a single aforementioned disclosed embodiment.

[0117] Each patent, patent application, publication of patent application, and other materials, such as articles, books, specifications, publications, documents, documents, etc., cited in this disclosure (excluding any historical review documents related thereto) are hereby incorporated by reference for all purposes related to this disclosure, such as in the specification and claims of this disclosure. However, if there is any inconsistency or conflict between the descriptions, definitions, and / or terminology of the above materials and the descriptions, definitions, and / or terminology used in this disclosure, the descriptions, definitions, and / or terminology used in this disclosure shall prevail.

[0118] Finally, it should be understood that the embodiments of the application disclosed herein are explanations of the principles of the embodiments of this specification. Other modified embodiments are also within the scope of this specification. Therefore, the embodiments disclosed in this specification are only used as examples and not as limitations. Those skilled in the art can adopt alternative configurations according to the embodiments in this specification to implement the applications in this specification. Therefore, the embodiments of this specification are not limited to the embodiments accurately described in the application.

Claims

1. A data processing method, comprising: Obtaining historical behavior data corresponding to a target user group, wherein the target user group includes a plurality of users known to have a preset behavior pattern; Generate M initial behavior sequences based on the historical behavior data, wherein each initial behavior sequence represents a plurality of behaviors performed by a user in the target user group in chronological order, and M is an integer greater than 1; Taking the M initial behavior sequences as an initial population, multiplying the initial population for multiple generations by simulating a population evolution process to obtain a target population, wherein the target population includes N target behavior sequences, and the N target behavior sequences have a higher ability to represent the preset behavior pattern than the M initial behavior sequences, and N is an integer greater than 1; and Based on each target behavior sequence in the target population, an identification feature corresponding to the preset behavior pattern is determined.

2. The method according to claim 1, wherein: The i-th generation reproduction process in the multi-generation reproduction includes: Selecting some behavior sequences that meet the evolution condition from the current population to form a first behavior sequence set, wherein the evolution condition is a condition related to the ability to represent the preset behavior pattern; performing a crossover operation on different behavior sequences in the first behavior sequence set to obtain a plurality of evolved behavior sequences, and adding the plurality of evolved behavior sequences to the first behavior sequence set to obtain a second behavior sequence set, and A population after the i-th generation of reproduction is generated based on the second behavior sequence set, wherein i is any natural number, when i=1, the current population is the initial population, and when i>1, the current population is the population after the i-1-th generation of reproduction.

3. The method according to claim 2, wherein: The evolving behavior sequence includes a plurality of subsequences, wherein the plurality of subsequences are respectively from at least two different behavior sequences in the first behavior sequence set.

4. The method according to claim 2, wherein: The step of generating the population after reproduction of the i-th generation based on the second behavior sequence set includes: performing mutation operations on some of the behavior sequences in the second behavior sequence set to obtain a plurality of mutated behavior sequences; and The multiple mutated behavior sequences and other behavior sequences in the second behavior sequence set that have not undergone the mutation operation are used as the population after the i-th generation reproduction.

5. The method according to claim 4, wherein: The mutation operation includes one or more of the following: Randomly add a preset number of behaviors to the behavior sequence; Randomly delete a preset number of behaviors in the behavior sequence; or, Randomly replace some behaviors in a behavior sequence with other behaviors.

6. The method according to claim 2, wherein: The step of selecting some behavior sequences satisfying the evolution condition from the current population to form a first behavior sequence set includes: Obtaining a fitness function corresponding to the preset behavior pattern, wherein the fitness function takes the behavior sequence as an independent variable and the fitness as a dependent variable, and the fitness obtained based on the fitness function represents the level of representation ability of a behavior sequence for the preset behavior pattern; Based on the fitness function, determining the fitness corresponding to each behavior sequence in the current population; Based on the fitness corresponding to each behavior sequence in the current population, some behavior sequences that meet the evolution condition are selected from the current population to form the first behavior sequence set.

7. The method according to claim 6, wherein: The fitness function includes a penalty term, which is configured to adjust the fitness of a behavior sequence based on the feasibility of the behavior sequence and satisfy the following constraints: When the feasibility of the first behavior sequence is lower than the feasibility of the second behavior sequence, the adjusted fitness corresponding to the first behavior sequence is lower than the adjusted fitness corresponding to the second behavior sequence.

8. The method according to claim 6, wherein: The selecting, based on the fitness corresponding to each behavior sequence in the current population, some behavior sequences that meet the evolution condition from the current population to form the first behavior sequence set includes: Determining a fitness mean corresponding to the current population based on the fitness corresponding to each behavior sequence in the current population; Determining the selection probability of each behavior sequence according to the ratio of the fitness mean to the fitness of each behavior sequence in the current population; and The behavior sequences whose selection probability is greater than a preset probability threshold are selected from the current population to form the first behavior sequence set.

9. The method according to claim 2, wherein: The termination condition of the multi-generation reproduction includes at least one of the following: The number of iterations that have been spawned is greater than or equal to the preset number of iterations, or The characterization capability of the behavior sequences included in the current population for the preset behavior pattern reaches a convergence condition.

10. The method according to claim 1, wherein: The generating M initial behavior sequences based on the historical behavior data comprises: Randomly obtain behavior data of M users from the historical behavior data; Obtaining a key behavior set corresponding to the preset behavior pattern, wherein the key behavior set includes a plurality of key behaviors related to the preset behavior pattern; and For each of the M users, multiple behaviors in the key behavior set are selected from the behavior data of the user, and an initial behavior sequence is generated based on the multiple behaviors.

11. The method according to claim 1, wherein: The method further comprises: Based on the identification features, users with the preset behavior pattern are mined from the user set.

12. A data processing system comprising: at least one storage medium storing at least one instruction set for performing a data processing method; as well as At least one processor is communicatively connected to the at least one storage medium, wherein the at least one processor reads the at least one instruction set when running, and performs the data processing method as described in any one of claims 1-11 according to the instructions of the at least one instruction set.

Images