Payment service pushing method, electronic equipment and computer readable storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and provides a payment service pushing method, electronic equipment and a computer readable storage medium, and the method comprises the steps that the electronic equipment starts to continuously collect the motion state of a user holding the electronic equipment from a preset time period before the predicted time of a payment behavior, and then responds to a predicted trigger event; and obtaining a motion state sequence of the user in a preset time period before the predicted trigger event. The electronic equipment performs characterization processing on the motion state sequence to obtain a motion state feature vector; and finally, inputting the motion state feature vector into a payment intention prediction model to obtain a prediction result of whether the user has the payment intention or not. And if the prediction result indicates that the user has the payment intention, the electronic equipment pushes a payment service card. Therefore, according to the method, the payment service card is pushed after the payment intention of the user is predicted and determined through the motion state of the user based on the regular motion before the payment behavior of the user, and the accuracy of pushing the payment service card can be improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of artificial intelligence technology, and in particular to a payment service push method, an electronic device, and a computer-readable storage medium. Background Art

[0002] With the rapid popularization of the Internet and the rapid development of electronic payment technology, electronic payment has become one of the most commonly used payment methods in users' daily lives. Most of the payment services in existing electronic devices are embedded in the corresponding payment applications, so users usually need to go through cumbersome payment operations to complete a payment.

[0003] In order to provide a more convenient payment process, simplify user payment operations and improve user experience, many manufacturers are trying to push payment service cards to users at a fixed time or location in advance based on the user's regular payment behavior. However, the traditional method of pushing payment service cards based on a fixed time or location has the problem of pushing them too far in advance or pushing them by mistake, which reduces the accuracy of payment service push. Summary of the invention

[0004] The embodiments of the present application provide a payment service push method, an electronic device, and a computer-readable storage medium, which are used to solve the problem of reduced accuracy in payment service card push.

[0005] To achieve the above objectives, the embodiments of the present application adopt the following technical solutions:

[0006] In a first aspect, a payment service push method is provided, the method comprising:

[0007] The electronic device continuously collects the motion state of the user holding the electronic device from a preset time period before the expected time of the payment behavior. The expected time of the payment behavior is the time when the user's historical applicable electronic device has made a payment behavior. Then, in response to the predicted trigger event, the electronic device obtains the user's motion state within the preset time period before the predicted trigger event to obtain a motion state sequence. The electronic device then characterizes the motion state sequence to obtain a motion state feature vector. Finally, the motion state feature vector is input into the payment intention prediction model to obtain a prediction result of whether the user has a payment intention. If the prediction result indicates that the user has a payment intention, the electronic device pushes the payment service card.

[0008] Because users usually move regularly to the payment location before making a payment. Therefore, the electronic device predicts the user's intention to pay based on the user's motion state before pushing the payment service card, thereby ensuring that the timing of the push is the time when the user may make a payment. It can be seen that compared to pushing payment service cards to users only at a fixed time or location, push based on motion state will be more accurate, thereby improving the accuracy of payment service card push. In a possible implementation of the first aspect, in order to improve the electronic device's understanding of the motion state and remove abnormal noise data in the sequence, the motion state sequence is characterized to obtain a motion state feature vector, which may include: mapping each motion state in the motion state sequence into a corresponding code to obtain a motion state coding sequence; smoothing the motion state coding sequence to obtain a motion state smoothed sequence; feature hashing the motion state smoothed sequence to obtain a motion state feature vector.

[0009] In a possible implementation of the first aspect, since the user's motion state usually does not change rapidly and continuously, a motion state will be maintained for a period of time, and therefore there may be redundant and unnecessary motion states in the motion state sequence. Therefore, each motion state in the motion state sequence is mapped to a corresponding code to obtain a motion state coding sequence, which may include: dividing the motion state sequence into multiple sequences at a preset time interval; taking the starting motion state of each sequence, and composing the codes corresponding to the starting motion states of the multiple sequences into a motion state coding sequence. It can be seen that based on the time interval, encoding the motion state sequence by segment can remove repeated motion states within a time period, thereby reducing redundant and unnecessary motion states in the sequence.

[0010] In a possible implementation of the first aspect, because the motion state sequence has large short-term noise and the movement of the position is also accidental, there are abnormal data such as burrs and noise in the sequence. Therefore, the smoothing process can be specifically smoothed by taking the median or the average value to smooth the burrs and other abnormal data. Based on this, the motion state coding sequence is smoothed to obtain a motion state smoothed sequence, which can include: traversing and updating the codes in the motion state coding sequence, the updated code is the average or median of the current traversal code and the preset number of neighboring codes before and after; the motion state coding sequence after the code update is used as the motion state smoothed sequence.

[0011] In a possible implementation of the first aspect, feature hashing is performed on a motion state smooth sequence to obtain a motion state feature vector, including: using a hash function to perform hash conversion on each code in the motion state smooth sequence to obtain a corresponding hash value, and mapping the hash value of each code to a dimensional index to determine the dimensional index of each code; superimposing the feature value on each dimensional index to obtain the motion state feature vector.

[0012] Among them, if the hash value of the code is greater than or equal to 0, the characteristic value -1 is superimposed on the dimension index corresponding to the code; if the hash value of the code is less than 0, the characteristic value 1 is superimposed on the dimension index corresponding to the code.

[0013] In a possible implementation of the first aspect, in order to improve the accuracy of payment intention prediction, while predicting payment intention based on the motion state feature vector, the probability that different types of motion state sequences actually correspond to payment behavior can be considered. Based on this, the payment service push method also includes: clustering the motion state smooth sequence with the motion sequence cluster, and taking the payment probability of the motion sequence cluster to which the motion state smooth sequence belongs as the user's payment behavior occurrence probability; wherein the motion sequence cluster is obtained by clustering the user's historical motion state sequence, and the payment probability of the motion sequence cluster is the proportion of the total number of historical motion state sequences corresponding to payment behavior in the cluster. Furthermore, inputting the motion state feature vector into the payment intention prediction model to obtain a prediction result of whether the user has payment intention can include: merging the probability of payment behavior with the motion state feature vector and inputting them into the payment intention prediction model to obtain a prediction result of whether the user has payment intention.

[0014] In a possible implementation of the first aspect, in order to improve the accuracy of payment intention prediction, in addition to motion state features, non-motion features and other indicators can also be combined to predict the user's payment intention, thereby predicting the user's payment intention in multiple dimensions and improving the accuracy of the prediction.

[0015] Based on this, the payment service push method may further include: obtaining non-motion state features within a preset time period before the prediction trigger event; wherein the non-motion state features include any one or more of time features, WIFI features, and elevator features. Therefore, inputting the motion state feature vector into the payment intention prediction model to obtain a prediction result of whether the user has a payment intention may include: merging the non-motion state features with the motion state feature vector and inputting them into the payment intention prediction model to obtain a prediction result of whether the user has a payment intention.

[0016] In a possible implementation of the first aspect, the prediction trigger event includes that the current time reaches the expected time of the payment behavior. Alternatively, in order to avoid the electronic device frequently acquiring the motion state sequence within the expected time of the payment behavior, thereby wasting resources and increasing power consumption, the prediction trigger event may also include that the current time reaches the expected time of the payment behavior and there is a return to desktop event.

[0017] Therefore, in response to a predicted trigger event, obtaining the user's motion state sequence within a preset time period before the predicted trigger event may include: when the current time reaches the expected time of the payment behavior, obtaining the user's motion state sequence within the preset time period before the current time; or, when the current time reaches the expected time of the payment behavior and there is a return to desktop event, obtaining the user's motion state sequence within the preset time period before the return to desktop event, thereby avoiding resource waste and increased power consumption.

[0018] In the second aspect, the present application provides an electronic device, including: one or more processors and a memory, the memory is coupled to the processor; one or more computer program codes are stored in the memory, and the computer program code includes computer instructions; when the processor executes the computer instructions, the electronic device performs the following steps: continuously collect the motion state of the user holding the electronic device from a preset time period before the expected time of the payment behavior; in response to a predicted trigger event, obtain the user's motion state within a preset time period before the predicted trigger event to obtain a motion state sequence; characterize the motion state sequence to obtain a motion state feature vector. The motion state feature vector is input into the payment intention prediction model to obtain a prediction result of whether the user has a payment intention. If the prediction result indicates that the user has a payment intention, push the payment service card.

[0019] In a possible implementation of the second aspect, when the above-mentioned computer instructions are executed by the processor, the electronic device also performs the following steps: mapping each motion state in the motion state sequence into a corresponding code to obtain a motion state code sequence; smoothing the motion state code sequence to obtain a motion state smoothed sequence; performing feature hashing on the motion state smoothed sequence to obtain a motion state feature vector.

[0020] In a possible implementation of the second aspect, when the above-mentioned computer instructions are executed by the processor, the electronic device further performs the following steps: dividing the motion state sequence into multiple sequences at preset time intervals; taking the starting motion state of each sequence, and combining the codes corresponding to the starting motion states of the multiple sequences into a motion state code sequence.

[0021] In a possible implementation of the second aspect, when the above-mentioned computer instructions are executed by the processor, the electronic device further performs the following steps: traversing and updating the codes in the motion state coding sequence, the updated code is the average or median of the current traversal code and a preset number of adjacent codes before and after; and using the motion state coding sequence after the code update as the motion state smoothing sequence.

[0022] In a possible implementation of the second aspect, when the above-mentioned computer instructions are executed by the processor, the electronic device further performs the following steps: using a hash function to perform hash conversion on each code in the motion state smooth sequence to obtain a corresponding hash value, and mapping the hash value of each code to a dimension index to determine the dimension index of each code; superimposing eigenvalues ​​on each dimensional index to obtain a motion state eigenvector; wherein, if the hash value of the code is greater than or equal to 0, superimposing a eigenvalue of -1 on the dimensional index corresponding to the code; if the hash value of the code is less than 0, superimposing a eigenvalue of 1 on the dimensional index corresponding to the code.

[0023] In a possible implementation of the second aspect, when the above-mentioned computer instructions are executed by the processor, the electronic device further performs the following steps: clustering the motion state smooth sequence with the motion sequence cluster, and taking the payment probability of the motion sequence cluster to which the motion state smooth sequence belongs as the probability of the user's payment behavior; wherein the motion sequence cluster is obtained by clustering the user's historical motion state sequence, and the payment probability of the motion sequence cluster is the proportion of the total number of historical motion state sequences with corresponding payment behaviors in the cluster; merging the payment behavior occurrence probability with the motion state feature vector and inputting them into the payment intention prediction model to obtain a prediction result of whether the user has payment intention.

[0024] In a possible implementation of the second aspect, when the above-mentioned computer instructions are executed by the processor, the electronic device further performs the following steps: obtaining non-motion state features within a preset time period before the predicted triggering event; wherein the non-motion state features include any one or more of time features, WIFI features, and elevator features; merging the non-motion state features with the motion state feature vector and inputting them into the payment intention prediction model to obtain a prediction result of whether the user has payment intention.

[0025] In a possible implementation of the second aspect, when the above-mentioned computer instructions are executed by the processor, the electronic device further performs the following steps: when the current time reaches the expected time of the payment behavior, obtaining the user's motion state sequence within a preset time period before the current time; or, when the current time reaches the expected time of the payment behavior and there is a return to desktop event, obtaining the user's motion state sequence within the preset time period before the return to desktop event.

[0026] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor in an electronic device, the electronic device executes the payment service push method of the first aspect and any possible design thereof.

[0027] In a fourth aspect, the present application provides a computer program product, which, when executed on a computer, enables the computer to execute the method of the first aspect and any possible design thereof. The computer may be the electronic device described above.

[0028] It can be understood that the beneficial effects that can be achieved by the electronic device of any possible design of the second aspect, the computer-readable storage medium of the third aspect, and the computer program product of the fourth aspect can be referred to as the beneficial effects in the first aspect and any possible design of the electronic device, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A schematic diagram of a payment service card display interface provided in an embodiment of the present application;

[0030] Figure 2 A schematic diagram of a code scanning interface provided in an embodiment of the present application;

[0031] Figure 3 A schematic diagram of the structure of an electronic device 100 provided in an embodiment of the present application;

[0032] Figure 4 A software structure block diagram of the electronic device 100 provided in an embodiment of the present application;

[0033] Figure 5 A flowchart of a payment service push method provided in an embodiment of the present application;

[0034] Figure 6 A schematic diagram of a motion state sequence provided in an embodiment of the present application;

[0035] Figure 7 Schematic diagram of the process of motion state sequence characterization processing provided in the embodiment of the present application Figure 1 ;

[0036] Figure 8 The principle framework of the model training and model application provided in the embodiment of the present application Figure 1 ;

[0037] Fig. 9 Schematic diagram of the process of motion state sequence characterization processing provided in the embodiment of the present application Figure 2 ;

[0038] Fig.10The principle framework of the model training and model application provided in the embodiment of the present application Figure 2 . DETAILED DESCRIPTION

[0039] The technical scheme of the embodiment of the present application will be clearly and completely described below in conjunction with the drawings in the embodiment of the present application. Among them, in the description of the embodiment of the present application, the terms used in the following embodiments are only for the purpose of describing a specific embodiment, and are not intended to limit the present application. In addition, in order to facilitate the clear description of the technical scheme of the embodiment of the present application, in the embodiment of the present application, if the words "first", "second" and the like are used to distinguish the same items or similar items with basically the same functions and effects. It can be understood by those skilled in the art that the words "first", "second" and the like do not limit the quantity and execution order, and the words "first", "second" and the like do not limit them to be different. And, in the description of the embodiment of the present application, unless otherwise specified, the meaning of "multiple" means two or more.

[0040] Electronic payment refers to the use of secure electronic means between users to securely transmit payment information to banks or corresponding processing institutions through information networks to realize the circulation of money (funds). At present, with the rapid popularization of the Internet and the rapid development of electronic payment technology, electronic payment has become one of the most commonly used payment methods in users' daily lives.

[0041] Payment services in existing electronic devices are mostly embedded in corresponding payment applications. Users usually need to go through cumbersome payment operations to complete a payment, including returning to the payment application interface, opening the payment application, opening the payment code, or opening the code scanning interface.

[0042] Therefore, in order to provide a more convenient payment process, simplify user payment operations and improve user experience, many manufacturers are trying to push payment service cards to users at a fixed time or location in advance based on the user's regular payment behavior. However, the traditional method of pushing payment service cards based on a fixed time or location has the problem of pushing them too far in advance or pushing them by mistake, which reduces the accuracy of payment service push.

[0043] In order to improve the accuracy of payment service push, an embodiment of the present application provides a payment service push method, which is applied to an electronic device. Generally speaking, users' payment behaviors usually occur at fixed payment locations, such as shops, business districts, etc. At the same time, when a user goes from one location to another payment location, the user's personal real-time state changes can be reflected by his motion state. Moreover, after analysis, it is found that there are usually regular changes in motion state before the user makes a payment behavior. Based on this, the main idea of ​​the payment service push method provided in the embodiment of the present application is: the electronic device calls the payment intention prediction model to predict whether the user has the intention to pay based on the user's motion state. After predicting that the user has the intention to pay, the electronic device pushes the payment service card to the user, thereby improving the accuracy of the payment service push.

[0044] Below, the payment service push method provided in the embodiment of the present application is described by taking an electronic device as an example.

[0045] The electronic device first continuously collects the motion state of the user holding the electronic device from a preset time period before the expected time of the payment behavior (for example, 30 minutes before the predicted trigger event). The expected time of the payment behavior can be the time when the user's historical use of the electronic device is expected to make a payment. Then, the electronic device waits for the predicted trigger event, and in response to the predicted trigger event, obtains the user's motion state sequence within the preset time period before the predicted trigger event. Among them, the motion state sequence is a sequence including multiple motion states, indicating the continuous motion of the user holding the electronic device within the preset time period. Therefore, the user's position regularity can be reflected by the motion state sequence, and the motion state included therein accompanies the user's position movement, and the electronic device can obtain it in real time. For example, a user goes from a fee payment location (such as a company) where there may be no payment behavior to another payment location (such as a business district, a shop) where there may be payment behavior. The motion state sequence to reach the payment location can be {stationary, stationary, walking, walking, running, running, walking, walking, stationary}.

[0046] Then, in order to facilitate the payment intention prediction model processing to improve the accuracy of the payment intention prediction model, the electronic device performs characterization processing on the acquired motion state sequence. The electronic device can convert the motion state sequence into a corresponding feature vector through the characterization processing of the motion state sequence, thereby obtaining a motion state feature vector. Exemplarily, the above motion state sequence can be converted into a 5-dimensional motion state feature vector {0, 3, 1, -4, 1} after characterization processing.

[0047] Finally, the electronic device inputs the motion state feature vector {0, 3, 1, -4, 1} obtained by characterization processing into the payment intention prediction model, and the payment intention prediction model obtains the prediction result of whether the user has payment intention based on the motion state feature vector, that is, the payment intention prediction model predicts and outputs the indication information of whether the user has payment intention under the motion state sequence. Among them, the payment intention prediction model is a model trained by the electronic device based on the historical motion state sequence corresponding to the user's payment behavior using the electronic device.

[0048] During the training process, the historical motion state sequence is also characterized and then input into the payment intention prediction model. Then, after training and learning the historical motion state sequence corresponding to the historical payment behavior, the payment intention prediction model can have the ability to predict the user's payment intention based on the motion state feature vector corresponding to the user's motion state sequence.

[0049] Exemplarily, the prediction result output by the payment intention prediction model may be a payment intention label or a payment intention probability used to indicate whether the user currently has payment intention. For example, the payment intention prediction model outputs a payment intention label "1" to indicate that the user may currently have payment intention, and outputs a payment intention label "0" to indicate that the user currently has no payment intention. Alternatively, the payment intention prediction model outputs a payment intention probability of 90%, indicating that the probability that the user currently has payment intention is 90%, and the electronic device can determine whether the user currently has payment intention based on the payment intention probability.

[0050] After the electronic device determines that the user may have a payment intention at the moment based on the prediction result output by the payment intention prediction model, it indicates that the user may currently make a payment behavior. Therefore, in order to reduce the user's payment operations, the electronic device directly pushes and displays the payment service card on the interface, thereby completing the push of the payment service card. Subsequently, the electronic device waits to receive the user's click operation on the payment service card, and directly displays the payment code or scan code interface in response to the click operation.

[0051] For example, Figure 1 and Figure 2 As shown, a schematic diagram of a payment service card display interface and a schematic diagram of a code scanning interface are provided. The electronic device can push the payment service card to be displayed at the top of the application interface, refer to Figure 1 When the electronic device receives the user's Figure 1 After clicking the payment service card shown, the electronic device displays the payment code interface, refer to Figure 2 .

[0052] In the embodiment of the present application, since the user usually has regular changes in motion state before making a payment to the payment location, compared with the traditional method of pushing payment services at a fixed time and location, the embodiment of the present application predicts the user's payment intention through the user's motion state, and can push the payment service card after accurately determining that the user has the payment intention, thereby improving the accuracy of the user's payment behavior prediction, and further ensuring the accuracy of the payment service card push.

[0053] In addition, after the electronic device starts collecting the motion state of the user holding the electronic device, in order to avoid the increase in power consumption caused by the electronic device continuously collecting. Therefore, in the embodiment of the present application, when the electronic device completes the push of the payment service card or the current time is after the expected time of the payment behavior, the electronic device can stop collecting the motion state of the user holding the electronic device. When the time reaches the preset time period before the expected time of the payment behavior next time, the electronic device starts collecting the motion state of the user holding the electronic device again.

[0054] For example, Figure 3 A schematic structural diagram of an electronic device 100 is provided.

[0055] The electronic device in the embodiment of the present application may include at least one of a mobile phone, a foldable electronic device, a tablet computer, a desktop computer, a laptop computer, a handheld computer, a notebook computer, an ultra-mobile personal computer (UMPC), a netbook, a cellular phone, a personal digital assistant (PDA), an augmented reality (AR) device, a virtual reality (VR) device, an artificial intelligence (AI) device, a wearable device, an in-vehicle device, a smart home device, or a smart city device. The embodiment of the present application does not impose any special restrictions on the specific type of the electronic device.

[0056] refer to Figure 3The electronic device 100 may include a processor 110, an external memory interface 120, an internal memory 121, a universal serial bus (USB) connector 130, a charging management module 140, a power management module 141, a battery 142, an antenna 1, an antenna 2, a mobile communication module 150, a wireless communication module 160, an audio module 170, a speaker 170A, a receiver 170B, a microphone 170C, an earphone interface 170D, a sensor module 180, a button 190, a motor 191, an indicator 192, a camera module 193, a display screen 194, and a subscriber identification module (SIM) card interface 195, etc. Among them, the sensor module 180 may include a pressure sensor 180A, a gyroscope sensor 180B, an air pressure sensor 180C, a magnetic sensor 180D, an acceleration sensor 180E, a distance sensor 180F, a proximity light sensor 180G, a fingerprint sensor 180H, a temperature sensor 180J, a touch sensor 180K, an ambient light sensor 180L, a bone conduction sensor 180M, etc. ( Figure 3 Only the gyro sensor 180B and the acceleration sensor 180E are shown).

[0057] The processor 110 may include one or more processing units, for example, the processor 110 may include an application processor (AP), a modem processor, a graphics processor (GPU), an image signal processor (ISP), a controller, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Among them, different processing units may be independent devices or integrated into one or more processors. The processor may generate an operation control signal according to the instruction opcode and the timing signal to complete the control of fetching and executing instructions.

[0058] A memory may also be provided in the processor 110 for storing instructions and data. For example, it may be used to store the payment intention prediction model of the embodiment of the present application and the motion state of the user holding the electronic device collected by the electronic device. In some embodiments, the memory in the processor 110 may be a cache memory. The memory may store instructions or data that have been used or are frequently used by the processor 110. If the processor 110 needs to use the instruction or data, it may be directly called from the memory. Repeated access is avoided, the waiting time of the processor 110 is reduced, and the efficiency of the system is improved.

[0059] Specifically to the embodiment of the present application, the payment intention prediction method can be executed by the processor 110 in the electronic device, and the processor 110 collects the motion state of the user holding the electronic device from a preset time period before the expected time of the payment behavior. Among them, the motion state is obtained according to the motion data collected by the sensor. For example, the processor 110 controls the gyroscope sensor 180B to collect angular velocity data and controls the acceleration sensor 180E to collect acceleration data, and then calculates the motion state according to the collected angular velocity data and acceleration data. Then, the processor 110 responds to the prediction trigger event and obtains the motion state sequence of the user in the preset time period before the prediction trigger event. It can be understood that the processor 110 obtains the motion state of the user in the preset time period before the prediction trigger event from the collected motion state of the user, and obtains the motion state sequence. Finally, the processor 110 performs characterization processing on the motion state sequence, inputs the obtained motion state feature vector into the payment intention prediction model, and obtains the prediction result of whether the user has the payment intention. After the prediction result indicates that the user has the payment intention, the processor 110 pushes the payment service card to be displayed on the display screen 194.

[0060] In some embodiments, the processor 110 may include one or more interfaces. The interface may include an inter-integrated circuit (I2C) interface, an inter-integrated circuit sound (I2S) interface, a pulse code modulation (PCM) interface, a universal asynchronous receiver / transmitter (UART) interface, a mobile industry processor interface (MIPI), a general-purpose input / output (GPIO) interface, a subscriber identity module (SIM) interface, and / or a universal serial bus (USB) interface, etc. The processor 110 may be connected to a touch sensor, an audio module, a wireless communication module, a display screen, a camera assembly, and the like through at least one of the above interfaces.

[0061] It is understandable that the interface connection relationship between the modules illustrated in the embodiment of the present application is only a schematic illustration and does not constitute a structural limitation on the electronic device 100. In other embodiments of the present application, the electronic device 100 may also adopt different interface connection methods in the above embodiments, or a combination of multiple interface connection methods.

[0062] The external memory interface 120 can be used to connect an external memory card, such as a Micro SD card, to expand the storage capacity of the electronic device 100. The external memory card communicates with the processor 110 through the external memory interface 120 to implement a data storage function. For example, files such as music and videos are stored in the external memory card. Or files such as music and videos are transferred from the electronic device to the external memory card.

[0063] The internal memory 121 can be used to store computer executable program codes, which include instructions. The internal memory 121 may include a program storage area and a data storage area. Among them, the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc. The data storage area may store data created during the use of the electronic device 100 (such as audio data, a phone book, etc.), etc. In addition, the internal memory 121 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, a universal flash storage (UFS), etc. The processor 110 executes various functional methods or data processing of the electronic device 100 by running instructions stored in the internal memory 121, and / or instructions stored in a memory provided in the processor.

[0064] The charging management module 140 is used to receive the charging input of the charger and charge the battery 142. While the charging management module 140 is charging the battery 142, it can also power the electronic device through the power management module 141.

[0065] The power management module 141 is used to connect the battery 142, the charging management module 140 and the processor 110. The power management module 141 receives input from the battery 142 and / or the charging management module 140, and supplies power to the processor 110, the internal memory 121, the display screen 194, the camera module 193, the wireless communication module 160, etc. The power management module 141 can also be used to monitor parameters such as battery capacity, battery cycle number, battery health status (leakage, impedance), etc. In some other embodiments, the power management module 141 can also be set in the processor 110. In other embodiments, the power management module 141 and the charging management module 140 can also be set in the same device.

[0066] The wireless communication function of the electronic device 100 can be implemented through antenna 1, antenna 2, mobile communication module 150, wireless communication module 160, modulation and demodulation processor and baseband processor. Among them, antenna 1 and antenna 2 are used to transmit and receive electromagnetic wave signals. The mobile communication module 150 can provide solutions for wireless communication including 2G / 3G / 4G / 5G applied to the electronic device 100. The wireless communication module 160 can provide solutions for wireless communication applied to the electronic device 100.

[0067] In some embodiments, antenna 1 of electronic device 100 is coupled to mobile communication module 150, and antenna 2 is coupled to wireless communication module 160, so that electronic device 100 can communicate with a network and other electronic devices through wireless communication technology.

[0068] The electronic device 100 can realize the display function through a GPU, a display screen 194, and an application processor. The GPU is a microprocessor for image processing, which is connected to the display screen 194 and the application processor. The GPU is used to perform mathematical and geometric calculations for graphics rendering. The processor 110 may include one or more GPUs, which execute program instructions to generate or change display information. The display screen 194 is used to display images, videos, etc.

[0069] The electronic device 100 can realize the camera function through the camera module 193, ISP, video codec, GPU, display screen 194, application processor AP, neural network processor NPU, etc.

[0070] It is to be understood that the structure illustrated in the embodiments of the present application does not constitute a specific limitation on the electronic device. In other embodiments of the present application, the electronic device may include more or fewer components than shown in the figure, or combine certain components, or split certain components, or arrange the components differently. The components shown in the figure may be implemented in hardware, software, or a combination of software and hardware.

[0071] refer to Figure 4 , which is a software structure block diagram of the electronic device 100 according to an embodiment of the present application.

[0072] The layered architecture divides the software into several layers, each with clear roles and division of labor. The layers communicate with each other through software interfaces. In some embodiments, the Android system is divided into four layers, from top to bottom: application layer (applications), application framework layer (application framework), hardware abstract layer (HAL) and kernel layer (kernel).

[0073] The application layer can include a range of applications.

[0074] like Figure 4 As shown, in the embodiment of the present application, the application layer includes at least perception applications, desktop applications, and payment applications. According to actual needs, the application layer may also include camera, gallery, calendar, call, map, navigation, WLAN, Bluetooth, music, video, short message and other applications, and the implementation of the present application does not impose any limitation on this.

[0075] The application framework layer provides an application programming interface (API) and a programming framework for the applications in the application layer. The application framework layer includes some predefined functions.

[0076] like Figure 4As shown, in the embodiment of the present application, the application framework layer includes a payment push service and a payment intention prediction model. The payment push service provides a push service for implementing the push of payment service cards. The payment intention prediction model is used to predict the user's payment intention.

[0077] The hardware abstraction layer runs in user space, encapsulates kernel layer drivers, and provides a calling interface to the upper layer. Figure 4 As shown, the hardware abstraction layer of the embodiment of the present application at least includes a sensor HAL. According to actual needs, the HAL layer may also include a display HAL, an audio HAL, a camera HAL, a Bluetooth HAL, etc., and the implementation of the present application does not impose any limitation on this.

[0078] Among them, the payment push service can interact with the sensor HAL in the HAL layer during operation. The sensor HAL is responsible for interacting with sensors (such as gyroscope sensors and acceleration sensors) in electronic devices that collect motion data (such as angular velocity data and acceleration data), thereby transmitting the motion data to the payment push service.

[0079] The kernel layer is the layer between hardware and software. Figure 4 As shown, the kernel layer of the embodiment of the present application at least includes a sensor driver. In addition, according to actual needs, the kernel layer may also include a display driver, a camera driver, an audio driver, a Bluetooth driver, etc., and the implementation of the present application does not impose any limitation on this.

[0080] Specific to the embodiments of this application, refer to Figure 4 On the one hand, the payment push service starts sending control instructions to the sensor HAL in a preset period of time before the expected time of the payment behavior, so that the sensor HAL calls the sensor driver in the kernel layer according to the received control instructions, and the sensor driver drives the sensor to collect motion data and report it. The payment push service determines the user's motion status based on the reported motion data.

[0081] On the other hand, the perception application responds to the predicted trigger event and calls the payment push service to push the payment service card. Based on the call of the perception application, the payment push service obtains the motion state within a preset time period before the predicted trigger time from the determined user motion state to obtain a motion state sequence. The motion state sequence is characterized to obtain a motion state feature vector, which is then input into the payment intention prediction model, and a prediction result indicating whether the user has a payment intention is obtained from the payment intention prediction model. Furthermore, the payment push service determines that the user has a payment intention based on the prediction result, and pushes the payment service card obtained from the payment application to the desktop application for display.

[0082] In some embodiments, Figure 5As shown, a flow chart of a payment service push method is provided, and the payment service push method is executed by an electronic device (such as a mobile phone), including steps S500-S504.

[0083] The following, combined Figure 5 The payment service push method of the embodiment of the present application is described.

[0084] S500, the electronic device continuously collects the motion status of the user holding the electronic device starting from a preset time period before the expected time of the payment behavior.

[0085] The estimated time of payment behavior is the estimated time when the user may use the electronic device to make a payment, which can be a precise time point or a time period. The electronic device can count the time when the user uses the electronic device to make a payment (for example, the time when the user opens the payment application on the electronic device and the time when the payment is completed through the payment application), analyze and determine the user's payment habits, and thus obtain the estimated time of the user's payment behavior.

[0086] For example, assuming that the user is usually accustomed to operating the payment application in the electronic device to complete the payment between 12:00 and 13:00 noon, the electronic device can determine that the estimated time of the user's payment behavior is 12:00-13:00. The electronic device begins to collect the motion state of the user holding the electronic device in a preset time period (for example, 30 minutes) before 12:00. In other words, the electronic device can start collecting the motion state of the user holding the electronic device from 11:30. Among them, since the user's motion state needs to be reflected based on the motion data of the electronic device, the motion state of the user holding the electronic device can be regarded as the user's motion state.

[0087] Alternatively, the electronic device can perform user profiling to obtain the estimated time of payment behavior. The electronic device can perform user profiling and analyze the user's historical behavior through relevant algorithms based on the acquired user data. At the same time, the user profile can predict the user's behavior or preferences that may occur in different scenarios based on the user's historical behavior. In an embodiment of the present application, the electronic device can count the user's payment behavior (such as the user's action and time to open the payment application, the time to complete the payment through the payment application, etc.) to generate a user profile. Then, the estimated time of the user's payment behavior is predicted based on the generated user portrait.

[0088] For example, an electronic device (such as a mobile phone) can generate a user profile by counting the user's actions and payment behaviors of operating a payment application within a preset time range. The preset time range can be any time length such as 1 week, 10 days, 15 days or a month.

[0089] In an embodiment of the present application, the motion state of the electronic device can be obtained by analyzing the position movement of the user. The electronic device calculates the user's moving speed based on the moving distance and moving time between the two positions after the user moves, and then determines the corresponding motion state according to the moving speed. Among them, the above-mentioned moving time can be a fixed time. The moving distance between the above-mentioned two positions can be the moving distance of the electronic device at a fixed time interval. That is, after the moving time of the user holding the electronic device reaches a certain time or the moving distance reaches a certain distance, the electronic device starts to calculate the user's moving speed within this time or this moving distance to determine the motion state of the user holding the electronic device.

[0090] According to different moving speeds, the user's motion state may include nine states, such as still, relative still, walking, micro-walking, running, elevator, subway, in-vehicle, and high-speed rail. Therefore, in the embodiment of the present application, the user's motion state sequence may include one or more of the above nine motion states.

[0091] It should be understood that, based on actual application requirements, the user's motion state may be increased or decreased accordingly according to actual requirements, and the embodiments of the present application do not limit this.

[0092] S501: In response to a predicted triggering event, the electronic device obtains a user's motion state sequence within a preset time period before the predicted triggering event.

[0093] In the embodiment of the present application, the prediction trigger event is a pre-set event used to trigger the electronic device to predict the user's payment intention. Among them, the prediction trigger event can be the current time reaching the expected time of the payment behavior. That is, when the current time reaches the expected time of the payment behavior, the electronic device can obtain the user's motion state sequence within the preset time period before the current time.

[0094] The motion state sequence is a sequence including multiple motion states. Each motion state in the motion state sequence corresponds to each time in the corresponding preset time period, indicating the continuous motion of the user holding the electronic device in the preset time period. In other words, from the perspective of time, each motion state in the motion state sequence is continuous from the head of the sequence to the end of the sequence.

[0095] Exemplarily, assume that the estimated time for payment is between 12:00 and 13:00 noon. When the electronic device detects that the current time is 12:00, the current time reaches the estimated time for payment, and the electronic device preliminarily determines that the user may have payment behavior. Therefore, the electronic device can further obtain the user's motion state sequence within a preset time period (for example, 30 minutes) before the current time 12:00, that is, in chronological order, obtain all the motion states of the user holding the electronic device within the 30 minutes from 11:30 to 12:00, thereby obtaining a continuous motion state sequence. Similarly, when the electronic device detects that the current time is 12:30, the electronic device obtains the continuous motion state of the user holding the electronic device within the 30 minutes from 12:00 to 12:30 in chronological order, thereby obtaining the user's motion state sequence within the time period from 12:00 to 12:30.

[0096] For example, Figure 6 As shown, the present application embodiment provides a schematic diagram of a motion state sequence. Figure 6 , when a user travels from one location to another payment location ( Figure 6 The figure can be understood as when going from a company building to a store. Based on the user's position change and moving speed, the motion state sequence acquired by the electronic device can be motion state sequence 1: {stationary, stationary, walking, walking, running, running, walking}. It can also be motion state sequence 2: {stationary, stationary, stationary, walking, walking, walking, walking}.

[0097] In some embodiments, in order to avoid the electronic device frequently obtaining the user's motion state sequence to predict the user's payment intention within the expected time of the payment behavior, thereby wasting electronic device resources and increasing power consumption. In addition, for payment scenarios that require displaying a payment code or scanning code interface, the user usually needs to return to the desktop to find the corresponding payment application before performing the payment operation. Therefore, the prediction trigger event can be further limited to an event of returning to the desktop within the expected time of the payment behavior.

[0098] That is, when the electronic device detects that the current time has reached the expected time for the payment behavior, and the user has an operation of returning to the desktop within the expected time for the payment behavior, the electronic device can determine that the user may have a payment behavior, and the electronic device obtains the user's motion state within a preset time period (for example, 30 minutes) before the event of returning to the desktop, and obtains a motion state sequence.

[0099] For example, assuming that the estimated time of the user's payment behavior is 12:00-13:00, and the electronic device detects that the user has returned to the desktop at the current time 12:30, the electronic device obtains the user's motion state sequence within a preset time period (for example, 30 minutes) before the corresponding time of the return to desktop event 12:30, that is, obtains the user's motion state within 30 minutes from 12:00 to 12:30, thereby obtaining the motion state sequence.

[0100] S502, the electronic device performs characterization processing on the motion state sequence to obtain a motion state feature vector.

[0101] Specifically, after the electronic device obtains the user's motion state sequence, in order to improve the accuracy of the payment intention prediction model in predicting the user's payment intention based on the motion state, the electronic device first converts the input data into a feature form that is better accepted by the model, so the electronic device first characterizes the motion state sequence to obtain the corresponding motion state feature vector.

[0102] In some embodiments, S502 may include: mapping each motion state in the motion state sequence into a corresponding code to obtain a motion state code sequence; smoothing the motion state code sequence to obtain a motion state smoothed sequence; performing feature hashing on the motion state smoothed sequence to obtain a motion state feature vector.

[0103] It can be seen that the embodiment of the present application performs characterization processing on the user's motion state sequence mainly including three steps: encoding mapping, smoothing processing and feature hash processing. Figure 7 , the present application embodiment provides a flow chart of motion state sequence characterization processing. Figure 7 These three steps are explained in detail.

[0104] Code mapping: The electronic device first performs code mapping on the motion state sequence, and maps each motion state in the motion state sequence to a corresponding code through code mapping. Among them, the code corresponding to each motion state can be set according to actual business needs, and the code corresponding to each motion state is in an increasing state based on the motion speed corresponding to the different motion states. The correspondence between the motion state and the code in the embodiment of the present application is shown in Table 1 below.

[0105] Table 1

[0106]

[0107]

[0108] Referring to Table 1 above, when the user's motion state sequence is {stationary, stationary, walking, walking, running, running, walking}, the motion state coding sequence obtained after the electronic device encodes and maps the motion state sequence is {1, 1, 2, 2, 3, 3, 2}. Alternatively, when the user's motion state sequence is {stationary, stationary, stationary, walking, walking, walking, walking}, the motion state coding sequence obtained after the electronic device encodes and maps the motion state sequence is {1, 1, 1, 2, 2, 2, 2}.

[0109] In some embodiments, the user's motion state usually does not change rapidly and continuously. That is, when the user is in a certain motion state, he or she usually stays in the motion state for a period of time. Therefore, the motion state coding sequence can be divided into segments according to time, and the coding mapping is performed based on the starting motion state of each segment, so that unnecessary redundant motion states can be removed.

[0110] Based on this, mapping each motion state in the motion state sequence into a corresponding code to obtain a motion state code sequence may include: dividing the motion state sequence into multiple sequences at preset time intervals; taking the starting motion state of each sequence, and combining the codes corresponding to the starting motion states of the multiple sequences into a motion state code sequence.

[0111] Specifically, when the electronic device encodes and maps the motion state sequence, the motion state sequence is discretized and encoded, and redundant motion states can be removed through discretized encoding. Specifically, discretized encoding means that the electronic device divides the motion state sequence into multiple sequences according to a preset time interval, and then encodes and maps each segment of the sequence as a unit.

[0112] For example, assume that the preset time interval is set to 1 minute according to actual needs, and assume that the preset time period corresponding to the motion state sequence is 30 minutes. Then, the electronic device divides the motion state sequence according to 1 minute intervals, and the motion state sequence can be divided into 30 segments. Then, the electronic device obtains the starting motion state of each segment of the sequence after segmentation, and composes the codes corresponding to each starting motion state into a motion state code sequence. If the motion state sequence is divided into 30 segments, the resulting motion state code sequence is a 1*30 vector.

[0113] For example, taking a short motion state sequence {stationary, stationary, stationary, walking, walking, walking, walking} as an example, when the electronic device segments the motion state sequence at 1-minute intervals, if the first motion state "stationary" and the second motion state "stationary" in the motion state sequence are within the first 1 minute, that is, these two stationary motion states will be segmented into the same segment sequence. Therefore, when the electronic device takes the code corresponding to the starting motion state in each segment to form a motion state code sequence, the motion state code sequence corresponding to the motion state sequence will be {1, 1, 2, 2, 2, 2}.

[0114] It can be seen that, compared with directly encoding and mapping the motion state sequence {stationary, stationary, stationary, walking, walking, walking, walking} to obtain a complete motion state coding sequence of {1, 1, 1, 2, 2, 2, 2}, the motion state coding sequence {1, 1, 2, 2, 2, 2} obtained by segmenting into multiple sequences and then encoding and mapping by segment obviously removes redundant motion states. It should be understood that the above-mentioned preset time interval of 1 minute is used in the example of the embodiment of the present application, and the specific preset time interval can be set according to actual business needs, which is not limited in the embodiment of the present application.

[0115] Smoothing: After the electronic device completes the code mapping to obtain the motion state code sequence, due to the large short-term noise in the motion state sequence and the randomness of the position movement, there are abnormal data such as glitches and noise in the sequence. Therefore, the electronic device further smoothes the motion state code sequence obtained by the code mapping to obtain a motion state smoothed sequence. The embodiment of the present application processes the motion state code sequence corresponding to the original motion state sequence in a smoothing manner, which can remove abnormal conditions such as glitches and noise in the sequence to improve the accuracy of subsequent predictions.

[0116] In some embodiments, the electronic device smoothes the motion state coding sequence to obtain a motion state smoothed sequence, which may include: traversing and updating the codes in the motion state coding sequence, where the updated codes are the average or median of the current traversed code and a preset number of adjacent codes before and after, and the motion state coding sequence after the updated codes is used as the motion state smoothed sequence.

[0117] Specifically, after the electronic device completes the motion state sequence coding, it traverses each code in the motion state code sequence and updates each code. The updated code is obtained by the electronic device by calculating the average value or median value of the current traversal code and the preset number of adjacent codes before and after it. The electronic device uses the calculated average value or determined median as the new value of the current traversal code.

[0118] For example, assuming that the preset number of configurations is 2, the electronic device can combine the first two adjacent codes and the last two adjacent codes of the current traversal code, and calculate the average value or median value of these four adjacent codes and the current traversal code itself, that is, a total of five codes. Then, the electronic device replaces the current traversal code with the determined average value or median as a new code.

[0119] For example, taking the motion state coding sequence {1, 1, 1, 2, 2, 2, 2} as an example, when the electronic device traverses to the fourth code 2, the fourth code is the current traversal code, and its first two adjacent codes are the second code 1 and the third code 1, and the last two adjacent codes are the fifth code 2 and the sixth code 2. Furthermore, if the electronic device takes the average value as the new code, then the new code is (1+1+2+2+2) / 5=1.6. The smooth motion state sequence obtained after the electronic device updates the code is {1, 1, 1, 1.6, 2, 2, 2}. If the electronic device takes the median as the new code, between the five codes "1, 1, 2, 2, 2", the median code is 2, then the smooth motion state sequence obtained after the electronic device is updated is {1, 1, 1, 2, 2, 2, 2}.

[0120] It should be understood that the above-mentioned preset number 2 is used as an example in the embodiment of the present application. The specific number in actual application can be set according to actual business needs, and the embodiment of the present application does not limit this.

[0121] Feature hashing: After the electronic device completes the smoothing of the sequence to obtain a motion state smoothing sequence, the electronic device constructs an embedding feature of the motion state sequence by performing feature hashing on the motion state smoothing sequence, thereby obtaining a motion state feature vector corresponding to the motion state sequence.

[0122] In some embodiments, feature hashing is performed on a motion state smooth sequence to obtain a motion state feature vector, including: using a hash function to perform hash conversion on each code in the motion state smooth sequence to obtain a corresponding hash value, and mapping the hash value of each code to a dimension index to determine the dimension index of each code; superimposing a feature value on each dimension index to obtain a motion state feature vector; wherein, if the hash value of the code is greater than or equal to 0, a feature value of -1 is superimposed on the dimension index corresponding to the code; if the hash value of the code is less than 0, a feature value of 1 is superimposed on the dimension index corresponding to the code.

[0123] Specifically, the electronic device uses a hash function, such as Murmurhash3, to perform hash conversion on the codes in the motion state smooth sequence to obtain the hash value corresponding to the code. Then, the electronic device maps the dimension index with the hash value of the code to determine the dimension index corresponding to each code. The mapping formula of the dimension index is as follows:

[0124] index=abs(hash(originalNumber))%n_features.

[0125] Where index represents the dimension index corresponding to the encoding originalNumber, abs represents the absolute value, hash(originalNumber) represents the hash value of the encoding originalNumber, and n_features represents the feature dimension after hash conversion. If n_features=5, it means that the motion state feature vector obtained after feature hashing is a 5-dimensional feature vector, then the value of the dimension index corresponding to each encoding is any value in [1, 2, 3, 4, 5]. If n_features=10, it means that the motion state feature vector obtained after feature hashing is a 10-dimensional feature vector, then the dimension index value of each encoding is any value in [1, 2, 3, 4, 5, 6, 7, 8, 9, 10].

[0126] It should be understood that the specific dimensional features of the motion state feature vector can be set according to actual needs. The embodiment of the present application is described below using 5-dimensional features as an example.

[0127] Therefore, after the electronic device obtains the dimensional index corresponding to each code, it determines the characteristic value (value) of each code according to the calculated hash (originalNumber), and then superimposes the corresponding characteristic value on the corresponding dimensional index. In an embodiment of the present application, the characteristic value of the code includes two values: 1 and -1. When hash (originalNumber) is greater than or equal to 0, the characteristic value of the code is 1. When hash (originalNumber) is less than 0, the characteristic value of the code is -1.

[0128] Exemplarily, before the eigenvalues ​​are superimposed, each eigenvalue in the 5-dimensional motion state eigenvector is 0, that is, {0, 0, 0, 0, 0}. If the hash value of the first code in the motion state smooth sequence is greater than or equal to 0, and the corresponding dimension index index = 4, the electronic device superimposes the eigenvalue 1 of the first code in the motion state smooth sequence on the dimension index 4, and the motion state eigenvector is updated to {0, 0, 0, 1, 0}. If the hash value of the second code in the motion state smooth sequence is less than 0, and the corresponding dimension index index = 3, the electronic device superimposes the eigenvalue -1 of the second code in the motion state smooth sequence on the dimension index 3, and the state eigenvector is updated to {0, 0, -1, 1, 0}. If the hash value of the third code in the motion state smooth sequence is less than 0, and the corresponding dimension index index = 3, the electronic device superimposes the eigenvalue -1 of the third code in the motion state smooth sequence on the dimension index 3, and the state eigenvector is updated to {0, 0, -2, 1, 0}. Similarly, other codes in the motion state smooth sequence are processed in the same way, which will not be repeated here.

[0129] When the electronic device superimposes the eigenvalues ​​corresponding to all the codes in the motion state smooth sequence onto the corresponding dimension index, the final motion state feature vector can be obtained. For example, the motion state feature vector can be Figure 7 The 5-dimensional features shown are {0, 3, 1, -4, 1}.

[0130] S503, the electronic device inputs the motion state feature vector into the payment intention prediction model to obtain a prediction result of whether the user has payment intention. If the prediction result indicates that the user has payment intention, the electronic device executes S504, and the electronic device pushes the payment service card. If the prediction result indicates that the user does not have payment intention, the electronic device returns to S500.

[0131] In the embodiment of the present application, the payment intention prediction model is a model that is trained with a historical motion state sequence corresponding to the user's payment behavior and has the ability to predict payment intention.

[0132] When the electronic device detects that the user has made a payment during the use of the user, the electronic device collects the user's motion state sequence in the preset time period before the payment behavior as model training data. That is, the electronic device collects the historical motion state sequence corresponding to the user's payment behavior as model training data. For example, when the electronic device detects that the user has completed a payment at 13:00, the electronic device collects the user's motion state sequence from 12:30 to 13:00 and records it as a model training data.

[0133] Then, the electronic device trains the payment intention prediction model based on all the collected model training data, thereby obtaining a payment intention prediction model with payment intention prediction capability. The model structure of the payment intention prediction model in the embodiment of the present application can adopt any existing model structure. For example, it can be an existing neural network model structure, or it can be a tree model structure such as GBDT, xgboost, etc.

[0134] It should be understood that during the model training process, the model training data also needs to be characterized before being input into the model. In other words, whether it is the model training phase or the model application prediction phase, the data input into the model needs to be processed in the same way. Figure 8 As shown, a principle block diagram of model training and model application is provided.

[0135] refer to Figure 8 , both the model training stage and the model application stage need to go through sequence characterization processing. In addition, in the model training stage, since model training is the learning stage of the model, the input model training data includes not only the historical motion state sequence, but also the payment intention label / probability corresponding to the historical motion state sequence. According to the historical motion state sequence and the corresponding payment intention label / probability, the payment intention is understood and learned to determine the optimal model parameters. After that, the model parameters are fixed to obtain a trained model.

[0136] In the model application stage, the payment intention is predicted using the trained model. Therefore, only the motion state sequence for which payment intention prediction is currently required needs to be input, and the model outputs the corresponding prediction result. The prediction result can be a payment intention label / probability based on the payment intention label / probability given in the model training stage. That is, the payment intention label is given in the model training stage, and the prediction result output in the model application stage is the payment intention label. Similarly, the payment intention probability is given in the model training stage, and the prediction result output in the model application stage is the payment intention probability.

[0137] Then, the electronic device can determine whether the user has the intention to pay at the moment based on the prediction results output by the payment intention prediction model. If the electronic device determines that the user has the intention to pay based on the prediction results output by the payment intention prediction model, the electronic device pushes the payment service card for display. Subsequently, the user can complete the payment by directly clicking on the payment service card to display the payment code or code scanning interface. That is, the electronic device waits for the user to click on the payment service card, and displays the payment code or code scanning interface in response to the click operation.

[0138] In some embodiments, the motion state sequences are classified according to whether the user actually has payment behavior, and the user's motion state sequences can be divided into motion state sequences corresponding to payment behaviors and motion state sequences corresponding to no payment behaviors. Therefore, for similar or identical motion state sequences, the proportion of motion state sequences with payment behaviors can be regarded as the payment probability that the user may make a payment under this type of motion state sequence.

[0139] Based on this, the electronic device can determine the probability of the user's payment behavior in the motion state sequence according to the type of the motion state sequence. Furthermore, when the electronic device predicts the user's payment intention based on the motion state characteristic vector corresponding to the motion state sequence, it can also predict the probability of the payment behavior corresponding to the motion state sequence, thereby further improving the accuracy of the payment prediction and ensuring the accuracy of the payment service push.

[0140] Therefore, the payment service push method may further include: clustering the motion state smooth sequence with the motion sequence cluster, and using the payment probability of the motion sequence cluster to which the motion state smooth sequence belongs as the probability of the user's payment behavior. The motion sequence cluster is obtained by clustering the user's historical motion state sequence, and the payment probability of the motion sequence cluster is the proportion of the total number of historical motion state sequences in the cluster that have corresponding payment behaviors.

[0141] Furthermore, the electronic device inputs the motion state feature vector into the payment intention prediction model to obtain a prediction result of whether the user has the payment intention, which may include: combining the probability of payment behavior with the motion state feature vector and inputting them into the payment intention prediction model, and the payment intention prediction model outputs a prediction result of whether the user has the payment intention based on the motion state feature vector combined with the probability of payment behavior. Fig. 9 The figure is a schematic diagram of the process of motion state sequence characterization processing provided in an embodiment of the present application.

[0142] refer to Fig. 9 In the motion state sequence characterization process, in addition to encoding mapping, smoothing processing and feature hashing processing, sequence similarity calculation can also be included. In the embodiment of the present application, the type of the current user's motion state sequence can be determined by sequence similarity calculation, and then the probability of the payment behavior can be determined.

[0143] Hereinafter, the sequence similarity calculation is described in detail in the embodiments of the present application.

[0144] Specifically, the electronic device collects the user's historical motion state sequences in advance and clusters the collected historical motion state sequences. The electronic device divides each historical motion state sequence into clusters through sequence clustering to obtain the motion sequence cluster of the embodiment of the present application. Among them, the historical motion state sequences that are divided into the same motion sequence cluster through clustering are similar or identical sequences, and each motion sequence cluster has a cluster center. Among them, the similarity calculation of different historical motion state sequences during clustering can adopt any existing similarity calculation method. For example, edit distance, Euclidean distance, cosine similarity, etc.

[0145] Then, the electronic device counts the total number of historical motion state sequences in each motion sequence cluster (hereinafter referred to as the number of sequences in the cluster) and the number of historical motion state sequences corresponding to payment behaviors (hereinafter referred to as the number of payment sequences in the cluster). The electronic device calculates the ratio of the number of payment sequences in the cluster to the number of sequences in the cluster to obtain the proportion of the number of payment sequences in the cluster to the total number. That is, the proportion of the total number = the number of payment sequences in the cluster / the number of sequences in the cluster. This proportion of the total number is regarded as the payment probability P of the motion sequence cluster in this embodiment of the application. The value range of the payment probability P is P∈[0,1].

[0146] Furthermore, in the process of processing payment service push, the electronic device can add a step of determining the payment behavior probability of the motion state sequence based on the predetermined payment probability of each motion sequence cluster in the process of motion state sequence characterization processing, that is, adding the processing of sequence similarity calculation.

[0147] Specifically, after the electronic device obtains the motion state smooth sequence corresponding to the motion state sequence through coding mapping and smoothing processing, it clusters the motion state smooth sequence by calculating the similarity between the motion state smooth sequence and each motion sequence cluster, thereby determining the motion sequence cluster to which the motion state smooth sequence belongs. Then, the electronic device uses the payment probability of the motion sequence cluster to which the motion state smooth sequence belongs as the probability of the user's payment behavior.

[0148] Among them, since the motion sequence cluster includes multiple motion state sequences, and the cluster center is the center of clustering, in order to achieve fast clustering while ensuring the accuracy of clustering of motion state smooth sequences, the electronic device can complete the division of motion state smooth sequences by calculating the similarity between the motion state smooth sequences and the cluster centers of each motion sequence cluster.

[0149] At the same time, in order to ensure that the two sequence data for similarity calculation are aligned, after the electronic device extracts the cluster center from the motion sequence cluster, it first performs encoding mapping and smoothing processing on the cluster center to obtain the corresponding smoothed sequence before performing similarity calculation. Of course, the electronic device can also complete this step in the clustering stage of the motion sequence cluster. That is, the electronic device performs encoding mapping and smoothing processing on all historical motion state sequences to obtain the corresponding smoothed sequence and then performs clustering to obtain the motion sequence cluster.

[0150] After the electronic device obtains the probability of the user's payment behavior, it concatenates and merges the probability of the payment behavior with the motion state feature vector. Fig. 9 , the probability of payment behavior "0.67" is combined with the 5-dimensional motion state feature vector {0, 3, 1, -4, 1} to obtain the final motion state feature vector {0, 3, 1, -4, 1, 0.67}. Subsequently, the electronic device inputs the combined motion state feature vector {0, 3, 1, -4, 1, 0.67} into the payment intention prediction model, and the payment intention prediction model outputs a prediction result based on the combined motion state feature vector {0, 3, 1, -4, 1, 0.67}, indicating whether the user has payment intention. After determining that the user has payment intention, the electronic device pushes the payment service card.

[0151] In addition, if it is found after clustering that the motion state smooth sequence does not belong to any motion sequence cluster, then the probability of the user's payment behavior is 0. For example, suppose the motion sequence cluster includes three clusters A, B, and C. The electronic device determines that the cluster to which the motion state smooth sequence belongs is A by calculating the similarity between the motion state smooth sequence and the cluster centers of the three clusters A, B, and C. Then the payment probability of cluster A is the probability of the user's payment behavior. If the electronic device determines that the current motion state smooth sequence does not belong to any cluster among A, B, and C, then the probability of the user's payment behavior is 0.

[0152] In some embodiments, in order to improve the accuracy of payment prediction, while predicting the user's payment intention based on the motion state feature vector, other non-motion state features can also be spliced ​​into the motion state feature vector and input into the payment intention prediction model for payment intention prediction. Optionally, since users usually make payments at fixed times, such as the corresponding time of three meals a day, holidays, weekends and other fixed times, the non-motion state features can include time features. Also, when the user goes to the payment location, the WIFI status of the electronic device will be disconnected. And the user is likely to need to take the elevator, so the non-motion state features can also include WIFI features and elevator features.

[0153] Based on this, the payment service push method may further include: obtaining non-motion state features within a preset time period before the prediction trigger event. Therefore, when the electronic device inputs the motion state feature vector into the payment intention prediction model and outputs the prediction result of whether the user has the payment intention, it may include: merging the non-motion state features with the motion state feature vector and inputting them into the payment intention prediction model to obtain the prediction result of whether the user has the payment intention.

[0154] Specifically, the electronic device may obtain the non-motion state characteristics of the user within the preset time period before the predicted triggering event while obtaining the motion state sequence of the user within the preset time period before the predicted triggering event.

[0155] For example, if a WIFI disconnection event occurs in the electronic device within 30 minutes before the predicted trigger event, a WIFI feature "1" is obtained; if a WIFI disconnection event does not occur in the electronic device, a WIFI feature "0" is obtained. Alternatively, if the electronic device determines that the user has taken an elevator by analyzing the position movement within 30 minutes before the predicted trigger event, an elevator feature "1" is obtained. If the electronic device determines that the user has not taken an elevator by analyzing the position movement, an elevator feature "0" is obtained.

[0156] Then, after the electronic device obtains the motion state feature vector according to the motion state sequence (the motion state feature vector can be a feature vector that incorporates the probability of payment behavior, or a feature vector that does not incorporate the probability of payment behavior), the electronic device merges the non-motion state feature with the motion state feature vector. Subsequently, the electronic device inputs the motion state feature vector that incorporates the non-motion state feature into the payment intention prediction model, and the payment intention prediction model outputs a prediction result of whether the user has payment intention based on the motion state feature vector that incorporates the non-motion state feature. When it is determined that the user has payment intention, the electronic device pushes and displays the payment service card.

[0157] For example, Fig.10 As shown, a principle block diagram of model training and model application with feature splicing is provided.

[0158] refer to Fig.10 In both the model training and model application stages, feature concatenation is added after the sequence characterization process. Through feature concatenation, the motion state feature vector output by the sequence characterization process is concatenated and merged with the probability of payment behavior and / or non-motion state features. Then, the motion state feature vector concatenated with the probability of payment behavior and / or non-motion state features is used as the input of the model.

[0159] In an embodiment of the present application, motion state features and non-motion state features are combined to predict the user's payment intention. The user's payment intention is predicted by considering multi-dimensional indicators, thereby improving the accuracy of the prediction.

[0160] Another embodiment of the present application provides an electronic device, which includes: a memory, a display screen and one or more processors, wherein the memory, the display screen and the processor are coupled; the display screen is used to display a payment service card, and the memory stores one or more computer program codes, and the computer program codes include computer instructions; when the processor executes the computer instructions, the electronic device implements the payment service push method described in any of the above embodiments.

[0161] Another embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor in an electronic device, the electronic device implements the payment service push method described in any of the above embodiments.

[0162] The embodiment of the present application also provides a computer program product. When the computer program product is run on a computer, the computer is enabled to execute each function or step in the above method embodiment.

[0163] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0164] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0165] The units described as separate components may or may not be physically separated, and the components shown as units may be one physical unit or multiple physical units, that is, they may be located in one place or distributed in multiple different places. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0166] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0167] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium, including several instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0168] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present application shall be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A payment service push method, characterized in that: include: Continuously collect the movement status of the user's electronic device from a preset time period before the estimated time of the payment behavior; The estimated time of payment behavior is the time when the user has made payment behavior using the electronic device in the past; In response to a predicted trigger event, obtaining a motion state sequence of the user within the preset time period before the predicted trigger event; wherein the motion state sequence includes a plurality of motion states, indicating the continuous motion of the user holding the electronic device within the preset time period; Performing characterization processing on the motion state sequence to obtain a motion state feature vector; Inputting the motion state feature vector into a payment intention prediction model to obtain a prediction result of whether the user has payment intention; the payment intention prediction model is trained based on a historical motion state sequence corresponding to the payment behavior; If the prediction result indicates that the user has payment intention, a payment service card is pushed.

2. The method according to claim 1, characterized in that The characterization process of the motion state sequence to obtain a motion state feature vector includes: Mapping each motion state in the motion state sequence into a corresponding code to obtain a motion state code sequence; Smoothing the motion state coding sequence to obtain a motion state smoothing sequence; Perform feature hashing processing on the motion state smooth sequence to obtain the motion state feature vector.

3. The method according to claim 2, characterized in that The step of mapping each motion state in the motion state sequence into a corresponding code to obtain a motion state code sequence comprises: Dividing the motion state sequence into multiple sequences at preset time intervals; The starting motion state of each sequence is taken, and the codes corresponding to the starting motion states of the multiple sequences are combined into the motion state code sequence.

4. The method according to claim 2 or 3, characterized in that: The step of smoothing the motion state coding sequence to obtain a motion state smooth sequence includes: Performing traversal updates on the codes in the motion state code sequence, wherein the updated code is an average value or median of the current traversal code and a preset number of adjacent codes before and after; The motion state coding sequence after coding update is used as the motion state smoothing sequence.

5. The method according to any one of claims 2 to 4, characterized in that: The performing feature hashing on the motion state smooth sequence to obtain the motion state feature vector includes: Using a hash function to perform hash conversion on each code in the motion state smooth sequence to obtain a corresponding hash value, and mapping the hash value of each code to a dimensional index to determine the dimensional index of each code; The eigenvalues ​​are superimposed on each of the dimensional indexes to obtain the motion state eigenvector; wherein, if the hash value of the code is greater than or equal to 0, the eigenvalue -1 is superimposed on the dimensional index corresponding to the code; if the hash value of the code is less than 0, the eigenvalue 1 is superimposed on the dimensional index corresponding to the code.

6. The method according to any one of claims 2 to 5, characterized in that: The method further comprises: Clustering the motion state smooth sequence with the motion sequence cluster, and taking the payment probability of the motion sequence cluster to which the motion state smooth sequence belongs as the payment behavior occurrence probability of the user; wherein the motion sequence cluster is obtained by clustering the historical motion state sequence of the user, and the payment probability of the motion sequence cluster is the proportion of the total number of historical motion state sequences corresponding to payment behavior in the cluster; The step of inputting the motion state feature vector into a payment intention prediction model to obtain a prediction result of whether the user has payment intention includes: The payment behavior occurrence probability is combined with the motion state feature vector and then input into the payment intention prediction model to obtain a prediction result of whether the user has payment intention.

7. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: Acquire non-motion state features within a preset time period before the predicted triggering event; wherein the non-motion state features include any one or more of time features, WIFI features, and elevator features; The step of inputting the motion state feature vector into a payment intention prediction model to obtain a prediction result of whether the user has payment intention includes: The non-motion state feature and the motion state feature vector are combined and input into the payment intention prediction model to obtain a prediction result of whether the user has payment intention.

8. The method according to claim 1, characterized in that The prediction trigger event includes that the current time reaches the estimated time of the payment behavior or the current time reaches the estimated time of the payment behavior and there is a return to desktop event; The step of obtaining, in response to a predicted triggering event, a sequence of motion states of the user within a preset time period before the predicted triggering event includes: When the current time reaches the estimated time of the payment behavior, obtaining the user's motion state sequence within the preset time period before the current time; Alternatively, when the current time reaches the estimated time of the payment behavior and there is a return to desktop event, the user's motion state sequence within the preset time period before the return to desktop event is obtained.

9. An electronic device, characterized in that: include: A memory, a display screen and one or more processors, wherein the memory, the display screen and the processor are coupled; The display screen is used to display a payment service card, and the memory stores one or more computer program codes, which include computer instructions; when the processor executes the computer instructions, the electronic device executes the payment service push method as described in any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor of an electronic device, the electronic device executes the payment service push method as described in any one of claims 1 to 8.

Images