A user behavior data processing method and device

Abstract

The present invention discloses a method and device for processing user behavior data, and relates to the field of computer technology. A specific implementation of the method includes receiving a user behavior log, obtaining data of each behavior type and the time when the corresponding behavior occurs; constructing user sample data based on a behavior model of time series information; wherein the behavior model based on time series information is a two-dimensional matrix of data of each behavior type and preset first time series data generated based on a sliding window; performing feature extraction on the user sample data through a preset convolutional neural network to determine that the user generates target behavior type data within a preset time period in the future, and then executing corresponding business procedures based on the target behavior type data. Therefore, the implementation of the present invention can solve the problems of incomplete capture of existing user behavior data and low data processing efficiency.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method and device for processing user behavior data. Background Art

[0002] In e-commerce scenarios, how to tap into the potential value of user behavior has always been an important technical direction for platforms. As for data, user behavior is often accompanied by changes over time, that is, time series. To capture this time series information, the traditional approach is to set a time window for sliding, thereby capturing rich time series information for modeling.

[0003] In the process of implementing the present invention, the inventors found that there are at least the following problems in the prior art:

[0004] At present, it is difficult to obtain comprehensive information based on sliding window time series capture, and when the number of sliding times is too large, it will interfere with the model. In addition, the setting of sliding window requires the assistance of complex feature engineering, which consumes too much manpower. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method and device for processing user behavior data, which can solve the problems of incomplete capture of existing user behavior data and low data processing efficiency.

[0006] To achieve the above-mentioned purpose, according to one aspect of an embodiment of the present invention, a user behavior data processing method is provided, including receiving a user behavior log, obtaining data of each behavior type and the time when the corresponding behavior occurs; constructing user sample data based on a behavior model of time series information; wherein the behavior model based on time series information is a two-dimensional matrix of data of each behavior type and preset first time series data generated based on a sliding window; performing feature extraction on the user sample data through a preset convolutional neural network to determine that the user generates target behavior type data within a preset time period in the future, and then executing a corresponding business program based on the target behavior type data.

[0007] Optionally, before constructing user sample data based on the behavior model of time series information, the following steps are included:

[0008] Obtain data of M types of behaviors in N unit time and generate an M*N two-dimensional matrix;

[0009] Through a sliding window with a window size of U and a step size of h, data of M types of behaviors are obtained to generate a two-dimensional matrix of M*L;

[0010] Concatenate the M*N two-dimensional matrix and the M*L two-dimensional matrix to generate an M*(L+N) behavior model based on timing information;

[0011] Among them, M is the data of each behavior type, N is the multiple of the unit time, and L is the difference between N unit time and the sliding window size U and the percentage calculated by the sliding window step length h plus 1.

[0012] Optionally, after extracting features from the user sample data using a preset convolutional neural network, the following steps are performed:

[0013] The extracted features are output to the maximum pooling layer and the average pooling layer respectively to obtain the first pooled data and the second pooled data;

[0014] Concatenate the first pooled data, the second pooled data, and the user behavior log.

[0015] Optionally, after splicing the first pooled data, the second pooled data, and the user behavior log, the process includes:

[0016] The concatenated data is input into the first fully connected network, and then into the Dropout layer network and the second fully connected network.

[0017] Optionally, after inputting into the Dropout layer network and the second fully connected network, it includes:

[0018] Based on the AUC scoring method, the output of the second fully connected network is optimized through the loss function.

[0019] Optionally, constructing user sample data based on a behavior model of time series information further includes:

[0020] The received user behavior logs are cleaned to construct user sample data based on the behavior model of time series information according to the cleaned user behavior logs.

[0021] Optionally include:

[0022] The RMSProp optimizer is used to pre-train the behavior model and convolutional neural network based on temporal information.

[0023] In addition, the present invention also provides a user behavior data processing device, including a receiving module, used to receive user behavior logs, obtain each behavior type data and the corresponding behavior occurrence time; construct user sample data based on the behavior model of time series information; wherein the behavior model based on time series information is a two-dimensional matrix of data of each behavior type and preset first time series data generated based on a sliding window; a processing module, used to extract features of the user sample data through a preset convolutional neural network to determine whether the user generates target behavior type data within a preset time period in the future, and then execute corresponding business procedures based on the target behavior type data.

[0024] An embodiment of the above invention has the following advantages or beneficial effects: because it adopts the method of receiving user behavior logs to obtain data of each behavior type and the time when the corresponding behavior occurs; constructing user sample data based on the behavior model of time series information; wherein the behavior model based on time series information is a two-dimensional matrix of data of each behavior type and preset first time series data generated based on a sliding window; performing feature extraction on user sample data through a preset convolutional neural network to determine that the user generates target behavior type data within a preset time period in the future, and then executing corresponding business procedures based on the target behavior type data, thereby overcoming the technical problems of incomplete capture of existing user behavior data and low data processing efficiency, thereby achieving the expansion of processing ideas on time series problems and realizing the technical effect of efficient user behavior data processing.

[0025] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0027] Figure 1 is a schematic diagram of the main flow of the method for processing user behavior data according to the first embodiment of the present invention;

[0028] Figure 2 is a schematic diagram of the main process of the user behavior data processing method according to the second embodiment of the present invention;

[0029] Figure 3 is a schematic diagram of main modules of a user behavior data processing device according to an embodiment of the present invention;

[0030] Figure 4 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0031] Figure 5 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0033] Figure 1is a schematic diagram of the main process of the method for processing user behavior data according to the first embodiment of the present invention. Figure 1 As shown, the user behavior data processing method includes:

[0034] Step S101, receiving user behavior logs, obtaining data of each behavior type and the corresponding behavior occurrence time; constructing user sample data based on the behavior model of time series information.

[0035] The behavior model based on time series information is a two-dimensional matrix of data of each behavior type and preset first time series data generated based on a sliding window.

[0036] Step S102, extracting features from user sample data through a preset convolutional neural network to determine whether the user generates target behavior type data within a preset time period in the future, and then executing corresponding business procedures based on the target behavior type data.

[0037] In some embodiments, the preset convolutional neural network can be a two-dimensional convolutional neural network.

[0038] B(i,j)=∑ m＝0 ∑ n＝0 k(m,n)*A(im,jn), where A is the convolved matrix, i.e. the constructed user sample data, and k(m,n) is the convolution kernel. Therefore, there is no need to construct complex features manually, and only simple statistical features are needed to achieve good results.

[0039] In addition, different local features can be obtained by convolution through different convolution kernels, so local features can be extracted by sliding the convolution kernel. At the same time, Z different convolution kernels can be set to extract richer local features, and Z is an adjustable hyperparameter.

[0040] In some other embodiments, the behavior model based on timing information in step S101 may be obtained through the following process:

[0041] Obtain data of M types of behaviors in N unit time and generate an M*N two-dimensional matrix. Obtain data of M types of behaviors through a sliding window with a window size of U and a step size of h to generate an M*L two-dimensional matrix. Concatenate the M*N two-dimensional matrix and the M*L two-dimensional matrix to generate an M*(L+N) behavior model based on timing information. Among them, M is the data of each behavior type, N is a multiple of the unit time, and L is the difference between N unit time and the sliding window size U and the percentage calculated by the sliding window step size h plus 1.

[0042] Specifically, the present invention takes each user as a sample and takes N unit time as the time window, so a one-dimensional matrix with a length of N can be obtained for each behavior of the user. For example, for click behavior, the number of clicks of the user in each of N days is counted, so clicks click = [1, 8, 0, 7, 9 ...] can be obtained. Similarly, other behaviors of users can be counted by day, such as browse browse = [4, 15, 3, 9, 13 ...], order order = [1, 3, 0, 4, 5 ...], etc. In addition, some interactive statistical features can be counted, such as interactive behaviors with commodities. Then these behavioral data are spliced ​​train = [click, browse, order, ...]. Assuming that M types of behavioral data are spliced, a two-dimensional matrix of M*N is obtained after splicing.

[0043] In addition, the number of user behaviors per day is counted here, with a sliding window of window size U and sliding step size h, to obtain a one-dimensional matrix of length L = (NU)% h+1, and after splicing according to the behavior type data, a two-dimensional matrix M*L is obtained.

[0044] Then, the two two-dimensional matrices obtained above are concatenated according to the horizontal axis to obtain a new two-dimensional matrix M*(L+N), which is a behavior model based on timing information, where N, U, and h are set as configurable hyperparameters.

[0045] For example: user behavior data on e-commerce platforms, such as clicks, shopping carts, favorites, orders, reviews, order amounts, and the time when the corresponding behaviors occurred. After obtaining the original data, the behavior model based on the time series information is used to construct samples. For example, by counting the number of behaviors that occurred within three days, we can get the number of clicks within three days, the number of times users added to the shopping cart within three days, the number of orders within three days, the number of favorites within three days, the order amounts within three days, etc. Then expand the time window to count the number of clicks within seven days, the number of times users added to the shopping cart within seven days, the number of times users placed orders within seven days, and the order amounts within seven days. At the same time, we can also count within 14 days and 20 days. Then, we use the statistical data as the input of the behavior model to predict whether the user will make a purchase in the next day, and then we can execute the corresponding business program (such as the item or advertisement push program) based on the predicted possible shopping behavior.

[0046] It is worth noting that when executing the behavior model based on time series information and constructing user sample data in step S101, the user behavior log can be received, so as to construct the user sample data based on the behavior model of time series information according to the cleaned user behavior log. Preferably, the original behavior log of the user of the platform is used as the original data, and then the original data is cleaned to remove abnormal data, missing data, etc.

[0047] In some other embodiments, step S102 outputs the features extracted from the user sample data by the preset convolutional neural network to the maximum pooling layer and the average pooling layer respectively to obtain the first pooled data and the second pooled data. Then, the first pooled data, the second pooled data and the user behavior log are spliced.

[0048] Furthermore, the spliced ​​data is input into the first fully connected network, and then into the Dropout layer network and the second fully connected network. Among them, passing through the Dropout layer network and the second fully connected network can prevent the model from overfitting. The first layer of the fully connected network integrates the pooling results and the user behavior log. The network structure of the first fully connected network and the second fully connected network is the same, and the parameters can be adjusted. The Dropout layer refers to the process of temporarily discarding the neural network unit from the network according to a certain probability during the training process of the deep learning network.

[0049] It is worth noting that based on the AUC scoring method, the output of the second fully connected network is passed through the loss function:

[0050]

[0051] Among them, i represents the i-th sample, m represents the total sample size, h(x) represents the predicted result, and y represents the actual result.

[0052] The target optimization is a binary classification problem, for example, whether the user will make a purchase in the next month, 1 for yes and 0 for no. AUC (Area Under Curve) is defined as the area under the ROC curve and the coordinate axis.

[0053] Of course, you can also set the optimization function according to different downstream tasks. For example, if the downstream is a regression task, the optimization function is the root mean square error function:

[0054]

[0055] Where n represents the number of samples, y i Indicates the actual result, y j Represents the prediction result.

[0056] As some further embodiments, the RMSProp (full name Root Mean Square prop) optimizer is used to pre-train the behavior model and convolutional neural network based on time series information. In addition, AUC is used as an evaluation index of the behavior model and convolutional neural network based on time series information. After training, it is verified in the divided verification set, and the hyperparameters are adjusted according to the verification results, such as the window size and the number of convolution kernels. Among them, the RMSProp (full name Root Mean Square prop) optimizer can solve the problem of excessive swing amplitude.

[0057] Preferably, 80% of the data is randomly divided as a training set and 20% as a validation set. The preset model is trained with 80% of the data, and the performance of the preset model is verified in the 20% data set.

[0058] In summary, the user behavior data processing method described in the present invention can solve the problem that the traditional capture of structured data time series information is not comprehensive and accurate enough, and realizes the universality of features through neural network technology, so that it can be applied to more than one downstream task, that is, it can execute corresponding downstream business procedures according to different target behavior type data in a preset time period in the future. In addition, it solves the problem that a large amount of complex feature engineering is required when modeling traditional structured data. At the same time, combined with the time window in time series related tasks, a new user sample construction method is proposed.

[0059] It can be seen that the present invention realizes the universality and complexity of features through neural networks, saves the time spent on manual feature replication engineering, and expands the processing ideas on time series problems.

[0060] Figure 2 is a schematic diagram of the main flow of a method for processing user behavior data according to a second embodiment of the present invention. The method for processing user behavior data may include:

[0061] Step S201, receiving user behavior log.

[0062] Step S202: construct user sample data based on the behavior model of time series information according to the cleaned user behavior log.

[0063] Step S203: extract features from the user sample data using a preset convolutional neural network.

[0064] Step S204: output the extracted features to the maximum pooling layer and the average pooling layer respectively to obtain the first pooled data and the second pooled data.

[0065] Step S205, concatenating the first pooled data, the second pooled data, and the user behavior log.

[0066] Step S206, input the concatenated data into the first fully connected network, and then input into the Dropout layer network and the second fully connected network.

[0067] Step S207: determining, based on the result output by the second fully connected network, whether the user generates target behavior type data within a preset time period in the future.

[0068] Figure 3 is a schematic diagram of the main modules of the user behavior data processing device according to an embodiment of the present invention. Figure 3 As shown, the user behavior data processing device 300 includes a receiving module 301 and a processing module 302. The receiving module 301 is used to receive the user behavior log, obtain each behavior type data and the corresponding behavior occurrence time; and construct user sample data based on the behavior model of the time series information. The processing module 302 is used to extract features from the user sample data through a preset convolutional neural network to determine whether the user generates target behavior type data within a preset time period in the future, and then execute the corresponding business program based on the target behavior type data.

[0069] The behavior model based on time series information is a two-dimensional matrix of data of each behavior type and preset first time series data generated based on a sliding window.

[0070] In some embodiments, before constructing the user sample data based on the behavior model of the timing information, the receiving module 301 obtains data of M types of behavior in N unit times to generate an M*N two-dimensional matrix. Through a sliding window with a window size of U and a step size of h, data of M types of behavior are obtained to generate an M*L two-dimensional matrix. The M*N two-dimensional matrix and the M*L two-dimensional matrix are concatenated to generate an M*(L+N) behavior model based on timing information; wherein M is the data of each behavior type, N is a multiple of the unit time, and L is the difference between N unit times and the sliding window size U and the percentage calculated by the sliding window step size h plus 1.

[0071] It is worth noting that the receiving module 301 constructs user sample data based on the behavior model of time series information, including receiving user behavior logs to construct user sample data based on the behavior model of time series information according to the cleaned user behavior logs.

[0072] As other embodiments, the processing module 302 outputs the extracted features to the maximum pooling layer and the average pooling layer respectively to obtain the first pooled data and the second pooled data; and splices the first pooled data, the second pooled data and the user behavior log.

[0073] In a further embodiment, the processing module 302 inputs the concatenated data into the first fully connected network, and then inputs the concatenated data into the Dropout layer network and the second fully connected network.

[0074] In a further embodiment, after the processing module 302 inputs the Dropout layer network and the second fully connected network, the result output by the second fully connected network is optimized by a loss function based on the AUC scoring method.

[0075] It is worth mentioning that the RMSProp optimizer is used to pre-train the behavior model and convolutional neural network based on temporal information.

[0076] It should be noted that the user behavior data processing method and the user behavior data processing device described in the present invention have a corresponding relationship in terms of specific implementation contents, so the repeated contents will not be described again.

[0077] Figure 4 An exemplary system architecture 400 is shown to which the user behavior data processing method or user behavior data processing device, user behavior data processing determination or user behavior data processing determination device according to the embodiments of the present invention can be applied.

[0078] like Figure 4 As shown, system architecture 400 may include terminal devices 401, 402, 403, network 404 and server 405. Network 404 is used to provide a medium for communication links between terminal devices 401, 402, 403 and server 405. Network 404 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0079] Users can use terminal devices 401, 402, 403 to interact with server 405 through network 404 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 401, 402, 403, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0080] Terminal devices 401, 402, 403 may be various electronic devices having a user behavior data processing screen or a user behavior data processing and identification screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0081] The server 405 may be a server that provides various services, such as a backend management server (only an example) that provides support for shopping websites browsed by users using the terminal devices 401, 402, and 403. The backend management server may analyze and process the received data such as product information query requests, and feed back the processing results (such as target push information, product information - only an example) to the terminal device.

[0082] It should be noted that the user behavior data processing method or user behavior data processing determination method provided in the embodiments of the present invention is generally executed by the server 405 , and accordingly, the computing device is generally disposed in the server 405 .

[0083] It should be understood that Figure 4 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0084] Reference below Figure 5 , which shows a schematic diagram of the structure of a computer system 500 of a terminal device suitable for implementing an embodiment of the present invention. Figure 5 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0085] like Figure 5 As shown, the computer system 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage part 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the computer system 500 are also stored. The CPU 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0086] The following components are connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including a cathode ray tube (CRT), a liquid crystal user behavior data processor (LCD), etc., and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed, so that a computer program read therefrom is installed into the storage section 508 as needed.

[0087] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 509, and / or installed from the removable medium 511. When the computer program is executed by the central processing unit (CPU) 501, the above-mentioned functions defined in the system of the present invention are executed.

[0088] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may 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 computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer 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. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0089] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0090] The modules involved in the embodiments of the present invention may be implemented by software or hardware. The modules described may also be set in a processor. For example, it may be described as follows: a processor includes a receiving module and a processing module. The names of these modules do not, in some cases, limit the modules themselves.

[0091] As another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs. When the above one or more programs are executed by a device, the device includes receiving a user behavior log, obtaining each behavior type data and the corresponding behavior occurrence time; constructing user sample data based on a behavior model of time series information; wherein the behavior model based on time series information is a two-dimensional matrix of each behavior type data and a preset first time series data generated based on a sliding window; performing feature extraction on the user sample data through a preset convolutional neural network to determine that the user generates target behavior type data within a preset time period in the future, and then executing the corresponding business program based on the target behavior type data.

[0092] The technical solution according to the embodiment of the present invention can solve the problems of incomplete capture of existing user behavior data and low data processing efficiency.

[0093] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for processing user behavior data, characterized in that: include: Receive user behavior logs, obtain data on each behavior type and the time when the corresponding behavior occurred; Constructing user sample data based on a behavior model of time series information; wherein the behavior model based on time series information is a two-dimensional matrix of data of each behavior type and preset first time series data generated based on a sliding window; The user sample data is subjected to feature extraction through a preset convolutional neural network to determine whether the user generates target behavior type data within a preset time period in the future, and then a corresponding business procedure is executed based on the target behavior type data; wherein, after the user sample data is subjected to feature extraction through a preset convolutional neural network, the method includes: outputting the extracted features to the maximum pooling layer and the average pooling layer, respectively, to obtain the first pooled data and the second pooled data; and splicing the first pooled data, the second pooled data and the user behavior log.

2. The method according to claim 1, characterized in that Before constructing user sample data based on the behavior model of time series information, the following steps are required: Obtain data of M types of behaviors in N unit time and generate an M*N two-dimensional matrix; Through a sliding window with a window size of U and a step size of h, data of M types of behaviors are obtained to generate a two-dimensional matrix of M*L; Concatenate the M*N two-dimensional matrix and the M*L two-dimensional matrix to generate an M*(L+N) behavior model based on timing information; Among them, M is the data of each behavior type, N is the multiple of the unit time, and L is the difference between N unit time and the sliding window size U and the percentage calculated by the sliding window step length h plus 1.

3. The method according to claim 1, characterized in that After splicing the first pooled data, the second pooled data and the user behavior log, it includes: The concatenated data is input into the first fully connected network, and then into the Dropout layer network and the second fully connected network.

4. The method according to claim 3, characterized in that After inputting into the Dropout layer network and the second fully connected network, it includes: Based on the AUC scoring method, the output of the second fully connected network is optimized through the loss function.

5. The method according to claim 1, characterized in that Based on the behavior model of time series information, construct user sample data, which also includes: The received user behavior logs are cleaned to construct user sample data based on the behavior model of time series information according to the cleaned user behavior logs.

6. The method according to any one of claims 1 to 5, characterized in that: include: The RMSProp optimizer is used to pre-train the behavior model and convolutional neural network based on temporal information.

7. A user behavior data processing device, characterized in that: The receiving module is used to receive user behavior logs and obtain data of each behavior type and the time when the corresponding behavior occurred; Constructing user sample data based on a behavior model of time series information; wherein the behavior model based on time series information is a two-dimensional matrix of data of each behavior type and preset first time series data generated based on a sliding window; A processing module is used to extract features from user sample data through a preset convolutional neural network to determine whether the user generates target behavior type data within a preset time period in the future, and then execute corresponding business procedures based on the target behavior type data; wherein, after extracting features from user sample data through a preset convolutional neural network, it includes: outputting the extracted features to the maximum pooling layer and the average pooling layer respectively to obtain first pooled data and second pooled data; and splicing the first pooled data, the second pooled data and the user behavior log.

8. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

9. A computer readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images