CN115239025A

CN115239025A - Payment prediction method and electronic equipment

Info

Publication number: CN115239025A
Application number: CN202211147427.5A
Authority: CN
Inventors: 赵杰
Original assignee: Honor Device Co Ltd
Current assignee: Honor Device Co Ltd
Priority date: 2022-09-21
Filing date: 2022-09-21
Publication date: 2022-10-25
Anticipated expiration: 2042-09-21
Also published as: CN115239025B

Abstract

The application provides a payment prediction method and electronic equipment, and relates to the technical field of artificial intelligence. According to the method, the payment event is regarded as an abnormal sample, the non-payment event is regarded as a normal sample, and the sample to be detected is judged to be the normal sample or the abnormal sample in an unsupervised abnormal detection mode, so that the data accumulation time can be shortened, and the personalized prediction of different users can be quickly realized. The method comprises the steps of obtaining a sample to be detected when the detection information triggers the abnormal detection condition, wherein the sample to be detected comprises user behavior data in a preset time period before the detection information triggers the abnormal detection condition; acquiring a normal sample set, wherein the normal sample set comprises a plurality of normal samples, and the normal samples comprise user behavior data in a preset time period before a non-payment event is detected; determining labels of samples to be tested based on the normal sample set, wherein the labels comprise payment events and non-payment events; and if the label of the sample to be detected is a payment event, pushing a payment service card.

Description

Payment prediction method and electronic equipment

Technical Field

The application relates to the technical field of artificial intelligence, in particular to a payment prediction method and electronic equipment.

Background

Most people are used to electronic payment mode in daily shopping. Considering that the electronic payment operation (including opening a payment application, opening a payment code, or opening a code scanning interface) is cumbersome, many manufacturers try to predict the payment intention of a user, so as to provide a convenient and fast payment process for the user.

Existing prediction methods primarily train the prediction model by accumulating payment data for the user and then using the payment data. The payment data may be user-specific or may be for multiple users. If the model is trained by using the payment data of a specific user, the obtained model can meet the individual requirements of the user and achieve the effect of accurate prediction, but the problem is that the payment data of a single user is sparse, a large amount of time is needed for collecting enough payment data, prediction cannot be performed during the process of collecting the payment data, and the experience of the user in using corresponding services is influenced. If the model is trained by using the payment data of a plurality of users, a large amount of payment data can be quickly collected, but the obtained model is difficult to meet the personalized requirements of the users, and the payment intention of each user cannot be accurately predicted.

It can be seen that the prior art has the problem that the data accumulation time which needs to wait for a long time is needed for accurately predicting the payment intention of an individual user.

Disclosure of Invention

The embodiment of the application provides a payment prediction method and electronic equipment, which can be used for rapidly accumulating data and accurately identifying the payment intention of an individual user.

In order to achieve the above purpose, the embodiment of the present application adopts the following technical solutions:

in a first aspect, a payment prediction method is provided, which is applied to an electronic device, and includes: when the detection information triggers the abnormal detection condition, acquiring a sample to be detected, wherein the sample to be detected comprises user behavior data in a preset time period before the detection information triggers the abnormal detection condition; acquiring a normal sample set, wherein the normal sample set comprises a plurality of normal samples, and the normal samples comprise user behavior data in a preset time period before a non-payment event is detected; determining labels of samples to be tested based on the normal sample set, wherein the labels comprise payment events and non-payment events; and if the label of the sample to be detected is a payment event, pushing a payment service card.

Therefore, the payment event is regarded as the abnormal sample, the non-payment event is regarded as the normal sample, and the sample to be detected is judged to be the normal sample or the abnormal sample by using the unsupervised abnormal detection mode, so that the data accumulation time can be shortened, the normal samples of different users can be collected quickly, and the personalized prediction of the different users can be realized.

In one implementation manner provided in the first aspect, determining the label of the sample to be tested based on the normal sample set includes: inputting a sample to be detected and a plurality of normal samples into a plurality of different anomaly detection models to obtain a plurality of first anomaly scores of each sample, wherein the anomaly detection models are obtained by training historical event samples of a user in at least one terminal device and corresponding labels, and the plurality of first anomaly scores are used for representing the difference degree of the two samples in different dimensions; normalizing the plurality of first anomaly scores of each sample; for each sample, carrying out weighted fusion on the multiple first abnormal scores subjected to normalization processing according to the weights of the multiple abnormal detection models to obtain a second abnormal score; sequencing the second abnormal score of each sample according to the sequence from large to small to obtain P samples with the maximum second abnormal score; and if the P samples comprise the samples to be detected, determining that the label of the samples to be detected is a payment event.

By setting different weights for different anomaly detection models, for example, if the accuracy of a certain anomaly detection model is higher, the corresponding weight is higher, and if the accuracy of a certain anomaly detection model is lower, the corresponding weight is lower, so that the finally obtained second anomaly score can reflect the difference degree between two samples more accurately. Meanwhile, the first abnormal score is normalized before fusion, so that errors caused by non-uniform dimensions among different first abnormal scores can be reduced.

In one implementation manner provided in the first aspect, determining the label of the sample to be tested based on the normal sample set includes: clustering a sample to be detected and a plurality of normal samples to obtain a plurality of sample clusters, wherein each sample cluster comprises a plurality of samples and a cluster center, and the plurality of samples comprise the sample to be detected and/or the normal samples; calculating the characteristic vector distance from each sample to the corresponding cluster center; obtaining N samples with the largest feature vector distance according to the feature vector distance arrangement sequence from large to small; and if the N samples comprise the samples to be detected, determining that the label of the samples to be detected is a payment event.

In one implementation manner provided by the first aspect, there are M sets of feature vectors between each sample and a corresponding cluster center, where each set of feature vectors includes a first feature vector of the sample and a second feature vector of the cluster center, and calculating a feature vector distance from each sample to the corresponding cluster center includes: calculating the feature vector distance between a first feature vector and a second feature vector included in each group of feature vectors in the M groups of feature vectors to obtain M first feature vector distances; normalizing the M first feature vector distances; and according to the weight of the M groups of eigenvectors, carrying out weighted fusion on the M first eigenvector distances subjected to the normalization processing to obtain the eigenvector distances.

Therefore, the clustering algorithm is improved according to actual services, different weights can be set for different features, and the obtained feature vector distance can reflect the similarity degree between two samples more accurately.

In one implementation manner provided in the first aspect, determining the label of the sample to be tested based on the normal sample set includes: inputting a sample to be detected and a plurality of normal samples into a plurality of different abnormal detection models to obtain a plurality of abnormal detection results, wherein the abnormal detection models are obtained by training based on historical event samples and corresponding labels of users in at least one terminal device; and if the number of the abnormal detection results of the payment events marked by the labels is larger than that of the abnormal detection results of the non-payment events marked by the labels, determining that the labels of the samples to be detected are the payment events.

In an implementation manner provided by the first aspect, determining the label of the sample to be tested based on the normal sample set includes: and inputting the sample to be detected and a plurality of normal samples into the abnormal detection model to obtain the label of the sample to be detected.

In one implementation manner provided by the first aspect, the anomaly detection model includes an anomaly detection model based on an isolated forest algorithm, an anomaly detection model based on a local anomaly factor algorithm, or an anomaly detection model based on a clustering algorithm.

In an implementation manner provided by the first aspect, the detecting information includes a current time, and when the detecting information triggers the abnormal detection condition, the obtaining the sample to be detected includes: and if the current moment is matched with the preset historical payment time, obtaining a sample to be tested, wherein the historical payment time is the time or the time period of the occurrence of the historical payment event. Therefore, the mobile phone can obtain the sample to be tested according to the payment habits (such as the habit payment time) of the user, and the effect of preliminarily predicting whether the user has the payment requirement is achieved.

In an implementation manner provided by the first aspect, the detecting information includes user location information, and when the detecting information triggers an abnormal detection condition, the obtaining the sample to be detected includes: and if the user position information is in a preset geographic fence, obtaining a sample to be detected, wherein the position center of the geographic fence is the position of the user when the historical payment event occurs.

In one implementation form provided in the first aspect, the method further includes: when a non-payment event is detected, acquiring a newly added normal sample; the newly added normal sample comprises user behavior data in a preset time period before a non-payment event is detected, and the non-payment event comprises events of displaying a desktop, starting a non-payment application, playing video/music and displaying a photo.

In one implementation form provided in the first aspect, the method further includes: and updating the normal sample set based on the newly-added normal samples according to a preset period. Therefore, the normal samples included in the normal sample set can be attached to the actual behaviors of the user, and the abnormality detection model is more personalized.

In one implementation form provided by the first aspect, the user behavior data includes data reflecting whether the user walks, runs or rides the vehicle, whether the user rides an elevator, whether WIFI is connected or disconnected, a time at which WIFI is connected or disconnected, a start time and an end time of walking, running or riding the vehicle, and a start time and an end time of riding the elevator.

In one implementation provided in the first aspect, the sample further comprises location data and environment data.

In a second aspect, the present application provides an electronic device comprising a processor, a memory coupled to the processor, the memory storing program instructions that, when executed by the processor, cause the electronic device to implement the method of any of the implementations of the first aspect.

In a third aspect, the present application provides a computer readable storage medium comprising computer instructions; the computer instructions, when executed on an electronic device, cause the electronic device to perform a method as any one of the implementations of the first aspect.

In a fourth aspect, there is provided a computer program product containing instructions which, when run on a computer, cause the computer to perform the method of quality of service monitoring of any one of the above first aspects.

In a fifth aspect, an apparatus (e.g., the apparatus may be a system-on-a-chip) is provided that includes a processor configured to enable a first device to implement the functionality referred to in the first aspect above. In one possible design, the apparatus further includes a memory for storing program instructions and data necessary for the electronic device. When the device is a chip system, the device may be formed by a chip, and may also include a chip and other discrete devices.

The technical effects brought by any one of the design manners in the second aspect to the fifth aspect may refer to the technical effects brought by different design manners in the first aspect, and are not described herein again.

Drawings

Fig. 1 is a schematic structural diagram of an electronic device 200 according to an embodiment of the present disclosure;

fig. 2 is a flowchart of a payment prediction method according to an embodiment of the present application;

fig. 3 is a flowchart of determining a label of a sample to be detected by a mobile phone according to an embodiment of the present disclosure;

fig. 4 is a flowchart of determining a label of a sample to be tested by using another mobile phone according to an embodiment of the present application;

fig. 5 is a schematic structural diagram of a chip system according to an embodiment of the present disclosure.

Detailed Description

In the following, the terms "first", "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present embodiment, "a plurality" means two or more unless otherwise specified.

The application mainly relates to a payment prediction method, which takes more non-payment events as normal events and fewer payment events as abnormal (abnormal) events, determines that a sample to be detected is a normal event or an abnormal event by comparing the sample to be detected with a plurality of non-payment event samples, and predicts that a payment demand exists in a user when the sample to be detected is determined to be an abnormal event. Due to the fact that the non-payment events are more frequent and the number of the non-payment events is larger than that of the payment events, the time required by the electronic equipment for accumulating enough normal samples can be reduced, and the effect of rapidly achieving the payment prediction function is achieved.

In the embodiment of the application, the payment event refers to an event that a user actively displays a payment code to pay or actively scans a collection code of a merchant to pay. A non-payment event may refer to an operation event performed by a user other than a payment event. For example, the non-payment event may refer to an event that the mobile phone returns to a desktop (which may also be referred to as a home interface, or the like), an event that the mobile phone opens a non-payment application in response to an operation of the user, or the like, and is not particularly limited herein.

Fig. 1 is a schematic structural diagram of an electronic device 200 according to an embodiment of the present disclosure.

As shown in fig. 1, the electronic device 200 may include: the mobile communication device includes a processor 210, an external memory interface 220, an internal memory 221, a Universal Serial Bus (USB) interface 230, a charging management module 240, a power management module 241, a battery 242, an antenna 1, an antenna 2, a mobile communication module 250, a wireless communication module 260, an audio module 270, a speaker 270A, a receiver 270B, a microphone 270C, an earphone interface 270D, a sensor module 280, keys 290, a motor 291, an indicator 292, a camera 293, a display 294, and a Subscriber Identity Module (SIM) card interface 295.

The sensor module 280 may include a pressure sensor, a gyroscope sensor, an air pressure sensor, a magnetic sensor, an acceleration sensor, a distance sensor, a proximity light sensor, a fingerprint sensor, a temperature sensor, a touch sensor, an ambient light sensor, a bone conduction sensor, and the like.

Processor 210 may include one or more processing units, such as: the processor 210 may include an Application Processor (AP), a modem processor, a Graphics Processor (GPU), an Image Signal Processor (ISP), a controller, a memory, a video codec, a Digital Signal Processor (DSP), a baseband processor, and/or a neural-Network Processing Unit (NPU), among others. Wherein, the different processing units may be independent devices or may be integrated in one or more processors.

The controller may be a neural center and a command center of the electronic device 200. The controller can generate an operation control signal according to the instruction operation code and the timing signal to complete the control of instruction fetching and instruction execution.

A memory may also be provided in processor 210 for storing instructions and data. In some embodiments, the memory in the processor 210 is a cache memory. The memory may hold instructions or data that have just been used or recycled by the processor 210. If the processor 210 needs to use the instruction or data again, it can be called directly from the memory. Avoiding repeated accesses reduces the latency of the processor 210, thereby increasing the efficiency of the system.

In some embodiments, processor 210 may include one or more interfaces. The interface may include an integrated circuit (I2C) interface, an integrated circuit built-in audio (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.

It should be understood that the connection relationship between the modules illustrated in the present embodiment is only an exemplary illustration, and does not limit the structure of the electronic device 200. In other embodiments, the electronic device 200 may also adopt different interface connection manners or a combination of multiple interface connection manners in the above embodiments.

The charge management module 240 is configured to receive a charging input from a charger. The charger may be a wireless charger or a wired charger. The charging management module 240 may also supply power to the terminal device through the power management module 241 while charging the battery 242.

The power management module 241 is used to connect the battery 242, the charging management module 240 and the processor 210. The power management module 241 receives input from the battery 242 and/or the charging management module 240, and provides power to the processor 210, the internal memory 221, the external memory, the display 294, the camera 293, and the wireless communication module 260. In some embodiments, the power management module 241 and the charging management module 240 may also be disposed in the same device.

The wireless communication function of the electronic device 200 may be implemented by the antenna 1, the antenna 2, the mobile communication module 250, the wireless communication module 260, the modem processor, the baseband processor, and the like. In some embodiments, antenna 1 of electronic device 200 is coupled to mobile communication module 250 and antenna 2 is coupled to wireless communication module 260, such that electronic device 200 may communicate with networks and other devices via wireless communication techniques.

The antennas 1 and 2 are used for transmitting and receiving electromagnetic wave signals. Each antenna in the electronic device 200 may be used to cover a single or multiple communication bands. Different antennas can also be multiplexed to improve the utilization of the antennas. For example: the antenna 1 may be multiplexed as a diversity antenna of a wireless local area network. In other embodiments, the antenna may be used in conjunction with a tuning switch.

The mobile communication module 250 may provide a solution including 2G/3G/4G/5G wireless communication applied on the electronic device 200. The mobile communication module 250 may include at least one filter, a switch, a power amplifier, a Low Noise Amplifier (LNA), and the like. The mobile communication module 250 may receive the electromagnetic wave from the antenna 1, filter, amplify, etc. the received electromagnetic wave, and transmit the electromagnetic wave to the modem processor for demodulation.

The mobile communication module 250 can also amplify the signal modulated by the modem processor, and convert the signal into electromagnetic wave through the antenna 1 to radiate the electromagnetic wave. In some embodiments, at least some of the functional modules of the mobile communication module 250 may be disposed in the processor 210. In some embodiments, at least some of the functional blocks of the mobile communication module 250 may be provided in the same device as at least some of the blocks of the processor 210.

The wireless communication module 260 may provide a solution for wireless communication applied to the electronic device 200, including WLAN (e.g., wireless fidelity, wi-Fi) network, bluetooth (BT), global Navigation Satellite System (GNSS), frequency Modulation (FM), near Field Communication (NFC), infrared (IR), and the like.

The wireless communication module 260 may be one or more devices integrating at least one communication processing module. The wireless communication module 260 receives electromagnetic waves via the antenna 2, performs frequency modulation and filtering processing on electromagnetic wave signals, and transmits the processed signals to the processor 210. The wireless communication module 260 may also receive a signal to be transmitted from the processor 210, frequency-modulate and amplify the signal, and convert the signal into electromagnetic waves via the antenna 2 to radiate the electromagnetic waves.

The electronic device 200 implements display functions via the GPU, the display screen 294, and the application processor. The GPU is a microprocessor for image processing, coupled to a display screen 294 and an application processor. The GPU is used to perform mathematical and geometric calculations for graphics rendering. Processor 210 may include one or more GPUs that execute program instructions to generate or alter display information.

The display screen 294 is used to display images, video, and the like. The display screen 294 includes a display panel.

The electronic device 200 may implement a shooting function through the ISP, the camera 293, the video codec, the GPU, the display screen 294, and the application processor. The ISP is used to process the data fed back by the camera 293. The camera 293 is used to capture still images or video. In some embodiments, electronic device 200 may include 1 or N cameras 293, N being a positive integer greater than 1.

The external memory interface 220 may be used to connect an external memory card, such as a Micro SD card, to extend the storage capability of the electronic device 200. The external memory card communicates with the processor 210 through the external memory interface 220 to implement a data storage function. For example, files such as music, video, etc. are saved in the external memory card.

Internal memory 221 may be used to store computer-executable program code, including instructions. The processor 210 executes various functional applications of the electronic device 200 and data processing by executing instructions stored in the internal memory 221. For example, in the present embodiment, the processor 210 may execute instructions stored in the internal memory 221, and the internal memory 221 may include a program storage area and a data storage area.

Wherein, the storage program area can store an operating system, an application program (such as a payment prediction function) required by at least one function, and the like. The storage data area may store data created or collected during use of the electronic device 200 (e.g., user behavior data, location data, weather data, etc.), and the like. In addition, the internal memory 221 may include a high-speed random access memory, and may further include a nonvolatile memory, such as at least one magnetic disk storage device, a flash memory device, a universal flash memory (UFS), and the like.

Electronic device 200 may implement audio functions via audio module 270, speaker 270A, receiver 270B, microphone 270C, headphone interface 270D, and an application processor, among other things. Such as music playing, recording, etc.

The keys 290 include a power-on key, a volume key, etc. The keys 290 may be mechanical keys. Or may be touch keys. The motor 291 may generate a vibration cue. The motor 291 can be used for both incoming call vibration prompting and touch vibration feedback. Indicator 292 may be an indicator light that may be used to indicate a state of charge, a change in charge, or may be used to indicate a message, missed call, notification, etc. The SIM card interface 295 is used to connect a SIM card. The SIM card can be attached to and detached from the electronic device 200 by being inserted into the SIM card interface 295 or being pulled out of the SIM card interface 295. The electronic device 200 may support 1 or N SIM card interfaces, N being a positive integer greater than 1. The SIM card interface 295 may support a Nano SIM card, a Micro SIM card, a SIM card, etc.

The electronic device 200 provided in the embodiment of the present application may be a mobile phone, a tablet computer, a computer with a wireless transceiving function, a Personal Communication Service (PCS) phone, a desktop computer, a Virtual Reality (VR) terminal device, an Augmented Reality (AR) terminal device, a wireless terminal in industrial control (industrial control), a wireless terminal in self driving (self driving), a wireless terminal in smart grid (smart grid), a wireless terminal in transportation safety (transportation safety), a wireless terminal in smart city (smart city), a wireless terminal in smart home (smart home), and the like, and is not particularly limited herein.

The methods in the following embodiments may be implemented in the electronic device 200 having the above-described hardware structure.

For the sake of understanding, the method provided by the embodiments of the present application will be described in detail below with reference to the accompanying drawings.

As shown in fig. 2, an embodiment of the present application provides a payment prediction method, which may be executed by an electronic device 200 (e.g., a mobile phone). The flow of the payment prediction method comprises steps S210-S240.

And S210, when the detection information triggers an abnormal detection condition, the mobile phone acquires a sample to be detected.

In embodiments of the present application, one sample includes data that can distinguish between payment events and non-payment events. For example, before a user makes a payment event, there is usually a series of associated behaviors, for example, before employee a eats a meal at noon, employee a needs to walk to an elevator entrance, take the elevator to floor 1, walk to a restaurant, and then perform a payment operation to purchase lunch, where walking and taking the elevator is the associated behavior of the payment event (i.e., employee a makes a payment). That is, when the user takes the elevator after walking for a period of time and then continues to walk, a payment event is initiated probably. As such, the sample may include user behavior data for a preset period of time prior to the user making a payment event or a non-payment event. The user behavior data includes data for reflecting user behaviors, such as data reflecting whether the user walks, runs or rides on a vehicle, start time and end time when the user walks, runs or rides on the vehicle, data reflecting whether the mobile phone is connected or disconnected with the WIFI, time when the mobile phone is connected or disconnected with the WIFI, data reflecting whether the user rides on an elevator, start time and end time when the user rides on the elevator, and the like.

Optionally, considering that weather and the location of the user may also become factors affecting the user initiating the payment operation, for example, the user may not go out to eat or shop on snow or rain, and thus the probability of the payment event is small; further alternatively, there is probably no shopping environment at home or in the office, and the probability of the payment event being present is small. Thus, a sample may also include location data and/or weather data for a preset period of time before a user makes a payment event or a non-payment event. The location data includes location information of the user, such as longitude and latitude, information of the province and city, district and county, street, and the like. The weather data is used to reflect weather conditions such as cloudy, sunny, snow, rain, etc. It can be seen that a sample may include user behavior data only, or user behavior data and location data, or user behavior data and weather data, or user behavior data, location data, and weather data. It should be noted that the sample may also include other data that can distinguish between payment events and non-payment events, and is not limited herein.

The preset time period before the payment event/non-payment event is performed may refer to a time period formed by using a time when the user performs the payment event/non-payment event as an initial point and a time when a preset time elapses as an end point. For example, the preset time may be half an hour, and if the mobile phone detects a non-payment event at 12.

The sample to be tested comprises user behavior data, position data and/or weather data in a preset time period before the detection information triggers the abnormal detection condition. When the mobile phone detection information triggers the abnormal detection condition, the payment requirement of the user can be preliminarily judged, so that the mobile phone further obtains user behavior data, position data and/or weather data in a preset time period before the detection information triggers the abnormal detection condition to form a sample to be detected, and whether the user can initiate payment operation or not is further accurately predicted through richer information.

The detection information may include information associated with the payment event. Optionally, the detection information includes user location information, which is used to reflect the current location of the user. The mobile phone can acquire the user position information in real time or according to a certain time interval. Correspondingly, the anomaly detection condition may indicate that the user location information is within a predetermined geo-fence. The preset geo-fence is constructed by extending the mobile phone outwards at a preset central position. The central location may be the location where the user was when the historical payment event occurred. Specifically, the mobile phone may analyze the location information of the user when the historical payment event occurs, and use the location information with a high frequency of occurrence as the center location. For example, the mobile phone may determine, by analyzing data of historical payment events, that a user mostly performs a payment operation at the a location, and then the mobile phone may extend outward from the a location to construct the geo-fence a. When the mobile phone detects that the user location information is in the geo-fence a, it is preliminarily determined that a payment event may exist, and user behavior data, location data and/or weather data within a preset time (for example, 20 minutes) before the user location information is detected in the geo-fence a are further acquired to form a sample to be tested.

Optionally, the detection information includes a current time. Correspondingly, the anomaly detection condition may refer to that the current time matches the historical payment time. The historical payment time is the time or time period of the historical payment event, and the current time matching the historical payment event may be understood as the current time being the same as the historical payment time or the current time being within the time period indicated by the historical payment time. Specifically, the mobile phone may analyze the occurrence time of the historical payment event, and use the time or the time period with a higher occurrence frequency as the historical payment time. For example, the mobile phone determines by analyzing the occurrence time of the historical payment event, and the user a usually performs the payment operation between 12 to 13 at noon. Therefore, when the current time is 12.

Optionally, the detection information may include user location information and current time, and correspondingly, the anomaly detection condition may indicate that the user location information is within a preset geo-fence and that the current time matches the historical payment time. That is, when the mobile phone detects that the user position information is in the preset geo-fence and the current time is matched with the historical payment time, the sample to be tested is obtained.

The detection information further includes other information associated with the payment event, which is not described in detail herein. By acquiring the sample to be detected when the detection information triggers the abnormal detection condition, preliminary screening can be performed, the frequency of acquiring/collecting data by the mobile phone is reduced, the frequency of performing abnormal detection by the mobile phone is further reduced, and the effect of reducing the energy consumption of the mobile phone is achieved.

Optionally, the mobile phone may further obtain the sample to be detected according to a preset time interval. For example, the mobile phone acquires user behavior data, position data and/or weather data in a preset time period at intervals of 1 hour as a sample to be detected, and performs anomaly detection on the sample to be detected.

S220, the mobile phone obtains a normal sample set.

Wherein the normal sample set includes a plurality of normal samples. The normal sample is a sample of historical non-payment events. The normal sample can be collected and stored by the mobile phone in the normal use process of the user. The normal sample may be used to reflect user behavior, actions prior to performing the non-payment event, including user behavior data, location data, and/or environmental data within a preset time period prior to detecting the non-payment event.

Non-payment events may include other events besides payment events, such as events that display a desktop (also may be referred to as a main interface, home interface), open non-payment applications, play video/music, display photos, and so forth.

In this embodiment of the application, when detecting that the desktop is displayed, the mobile phone may obtain, as a normal sample, user behavior data, location data, and/or environment data within a first preset time period, where the first preset time period is a time period formed by taking a time when the desktop is displayed by the mobile phone as a starting point and taking a time when the starting point is forward by a preset time (for example, half an hour) as an end point. Alternatively, the cell phone may also obtain normal samples when other non-payment events are detected.

Optionally, the mobile phone may further collect normal samples according to a preset period. Specifically, the mobile phone can detect non-payment events in the current period, acquire user behavior data, position data and/or environment data in preset time before each payment event occurs, serve as a sample corresponding to the payment event, and serve as a normal sample.

It will be appreciated that as non-payment events are more numerous and frequent in daily life relative to payment events, the handset can accumulate a large number of normal samples for anomaly detection at a faster rate.

And S230, determining the label of the sample to be detected based on the normal sample set by the mobile phone.

Wherein the tag includes a payment event and a non-payment event.

The ways for determining the label of the sample to be detected based on the normal sample set by the mobile phone include various ways, which will be described below.

In the first mode, the mobile phone can determine the label of the sample to be detected by using a single anomaly detection model.

Specifically, the mobile phone can input the sample to be detected and the normal sample set to the abnormal detection model together to obtain the label of the sample to be detected. The anomaly detection model can be trained by a mobile phone or other electronic devices (such as a PC and a server) by using historical event samples (including payment event samples and non-payment event samples) of a user in at least one terminal device and corresponding labels.

The anomaly detection model may be an anomaly detection model based on an independent forest (IForest) algorithm, an anomaly detection model based on a local anomaly factor (LOF) algorithm, or an anomaly detection model based on a clustering algorithm, and the like, which is not limited herein.

And secondly, the mobile phone determines the label of the sample to be detected by using a clustering algorithm improved based on the service condition.

For example, the mobile phone may determine the label of the sample to be detected based on the normal sample set by using the flow shown in fig. 3. The process includes S310-S340.

S310, clustering the to-be-detected sample and the plurality of normal samples by the mobile phone to obtain a plurality of sample clusters.

Wherein each sample cluster comprises a plurality of samples, which may include samples to be tested and/or non-payment event samples, and a cluster center.

Specifically, the mobile phone can randomly take K samples from the sample to be measured and a plurality of normal samples as cluster centers, then calculate the eigenvector distance from each sample to the K cluster centers, and divide the sample into the sample cluster where the cluster center with the smallest eigenvector distance from the sample is located, so as to obtain K sample clusters. Since the aforementioned cluster center is randomly drawn by the handset, it is not necessarily a true cluster center. In order to divide the more similar samples into the same sample cluster, the cell phone may re-determine the cluster center of each sample cluster. For example, the mobile phone may take an arithmetic mean of samples in each cluster, and re-cluster the samples with the arithmetic mean as a new cluster center until the result obtained by clustering no longer changes, and output K groups of sample clusters and the cluster center of each sample cluster.

Wherein the feature vector distance between two samples can be used to reflect the similarity of the two samples. If the distance of the feature vector between the two samples is larger, the similarity of the two samples is smaller; the smaller the feature vector distance between two samples, the greater the degree of similarity between the two samples.

The process of calculating the feature vector distance from the sample to the cluster center by the mobile phone may specifically be: the mobile phone extracts the characteristic information of the sample and the characteristic information of the cluster center, and then calculates the characteristic vector distance between the characteristic information of the sample and the characteristic information of the cluster center. Specifically, the sample and the cluster center include features having the same dimension, for example, both include M-dimensional features. Thus, there may be M sets of feature vectors between the sample and the cluster center, each set of feature vectors including a first feature vector for the sample and a second feature vector for the cluster center. The mobile phone can calculate the eigenvector distance between the first eigenvector and the second eigenvector included in each group of eigenvectors in the M groups of eigenvectors to obtain M first eigenvector distances. The mobile phone can also perform normalization processing on the M first feature vector distances so as to avoid errors caused by non-uniform dimensions among the features. The mobile phone can also perform weighted fusion on the M first eigenvector distances after normalization processing according to the weights of the M groups of eigenvectors to obtain the eigenvector distances.

That is, the feature vector distance between two samples (e.g., sample 1 and sample 2) can be calculated by equation 1.

Wherein, D is the distance of the feature vector, x1i is the ith dimension feature of the sample 1, x2i is the ith dimension feature of the sample 2, and wi is the weight of the ith dimension feature. It should be noted that, in equation 1, the eigenvector distance is exemplified as the euclidean distance, and the eigenvector distance may also be other distances, such as manhattan distance, which is not limited herein.

The mobile phone can set different weights for different characteristics according to actual service requirements. For example, the weight of more important features is set higher and the weight of less important features is set lower. The feature vector distance obtained in this way can reflect the similarity degree between two samples more accurately.

S320, the mobile phone determines a feature vector distance from each sample to the corresponding cluster center.

The mobile phone may calculate a feature vector distance from each sample to a corresponding cluster center based on equation 1, or directly obtain the feature vector distance calculated in the clustering process.

S330, the mobile phone sequences the distances of the plurality of eigenvectors according to the sequence from big to small to obtain N samples with the largest distance of the eigenvectors.

The N samples with the largest distance of the characteristic vectors are abnormal samples screened by the mobile phone, namely samples of payment events. In an optional embodiment, the mobile phone may preset an abnormal sample proportion, and N may be determined by the total number of samples (the sum of the numbers of all samples in the set of samples to be measured and normal samples) and the abnormal sample proportion. Specifically, N is the product of the total number of samples and the proportion of abnormal samples. For example, if the proportion of the abnormal samples is 1%, only the sample with the largest distance between feature vectors among 100 samples is an abnormal sample. For another example, when the proportion of the abnormal samples is 0.2% and the total number of the samples is 10000, the mobile phone can obtain 20 abnormal samples (i.e., N = 20).

And S340, if the N samples comprise the samples to be detected, the mobile phone determines that the label of the samples to be detected is a payment event.

And in the third mode, the mobile phone determines the label of the sample to be detected by using the plurality of abnormality detection models.

For example, the mobile phone may determine the label of the sample to be detected by using a plurality of anomaly detection models by using the flow shown in fig. 4. The process includes S410-S450.

S410, the mobile phone inputs the sample to be detected and the normal samples into a plurality of different anomaly detection models to obtain a plurality of first anomaly scores of each sample.

The plurality of different anomaly detection models may be an IForest-based anomaly detection model, an LOF-based anomaly detection model, a clustering algorithm-based anomaly detection model, and the like. The first anomaly scores output by the different anomaly detection models are used for representing the difference degree of the two samples in different dimensions. Where the different dimensions include density, distance, etc.

Each anomaly detection model can output a first anomaly score of the sample to be detected and a plurality of normal samples. For example, the mobile phone inputs 1 sample to be tested and 1000 normal samples to 3 different abnormality detection models together, and each abnormality detection model can input the first abnormality score of 1001 samples (including 1 sample to be tested and 1000 normal samples). Thus, each sample corresponds to a plurality of first anomaly scores.

S420, the mobile phone performs normalization processing on the plurality of first anomaly scores of each sample.

The error caused by the non-uniform dimension among different first anomaly scores can be reduced through the normalization processing.

And S430, weighting and fusing the plurality of normalized first abnormal scores by the mobile phone according to the weights of the plurality of abnormal detection models for each sample to obtain a second abnormal score.

Wherein the plurality of first anomaly scores and the plurality of second anomaly scores may satisfy equation 2.

Wherein S is ₂ Is the second anomaly score, ki is the weight of the ith anomaly detection model, S _1i A first anomaly score output for the ith anomaly detection model.

It should be noted that the mobile phone may calculate the second anomaly score of each sample by using equation 2.

By fusing the anomaly scores output by the anomaly detection models, the finally obtained second anomaly score can more accurately reflect the difference degree between the two samples.

The weights of the plurality of anomaly detection models may be set by a user according to the accuracy of each anomaly detection model, for example, if the accuracy of a certain anomaly detection model is higher, the corresponding weight is higher, and if the accuracy of a certain anomaly detection model is lower, the corresponding weight is lower.

Alternatively, the weights of the plurality of anomaly detection models may be determined by the cell phone or other electronic device through a grid parameter adjusting mechanism. Illustratively, the number of the abnormality detection models may be m, and assuming that the first m-1 abnormality detection models have weights of A1, A2 \8230 \8230am-1, the last abnormality detection model has a weight of 1-A1-A2- \8230: -Am-1. A1 and A2 (8230) (\ 8230), am-1 (0.05, 0.1, 0.15) (\ 8230), 0.30and A1+ A2+ \ 8230 (\ 8230) and Am-1 < 1) are set. The electronic device can cycle through the value range of each weight to determine multiple sets of weight coefficients. Then, the electronic device may fuse the first anomaly scores output by the plurality of anomaly detection models based on each set of weight coefficients, and finally select a set of weight coefficients with the most accurate fusion result. The value range may be other values, for example, 0.01,0.02,0.03, 8230, 0.99, 0.001,0.002,0.003, 8230, 0.999. The smaller the difference between two adjacent parameters in the value range is, the larger the grid density is, and the more accurate the finally obtained weight coefficient is.

Optionally, the weights of the abnormality detection models may be obtained by training a logistic regression model with the mobile phone or other electronic devices. The electronic equipment can take the abnormal scores output by the abnormal detection models as input, take the real label of the sample to be detected as the input label to train the model, and output the weights of the abnormal detection models until the model converges.

The two provided ways for determining the weights of the plurality of abnormal detection models are obtained by the electronic equipment according to the actual service situation, are more accurate than the weights set by the user subjectively, and can enable the fusion result of the plurality of abnormal detection models to be more practical.

S440, sorting the second abnormal score of each sample according to the sequence from large to small to obtain P samples with the maximum second abnormal score.

The P samples with the largest second abnormal score are abnormal samples screened by the mobile phone, namely samples of payment events. In an optional implementation manner, the mobile phone may preset an abnormal sample proportion, and P may be determined by the total number of samples (the sum of the number of the sample to be measured and the number of the plurality of normal samples) and the abnormal sample proportion. Specifically, P is the product of the total number of samples and the proportion of abnormal samples. For example, if the proportion of the abnormal samples is 0.1%, only the sample with the largest second abnormal score among the 1000 samples is the abnormal sample. For another example, when the proportion of the abnormal samples is 0.5% and the total number of the samples is 1000, the mobile phone may obtain 5 abnormal samples (i.e., P = 5).

S450, if the P samples comprise the samples to be detected, determining that the labels of the samples to be detected are payment events.

And if the second abnormal score is smaller than a preset first threshold value, the mobile phone determines that the label of the sample to be detected is a non-payment event.

Optionally, the plurality of anomaly detection models may further output a label of the sample to be detected. In this case, the mobile phone inputs the sample to be detected and the plurality of normal samples into a plurality of different anomaly detection models to obtain a plurality of anomaly detection results. The abnormal detection result comprises a label of the sample to be detected, and the label comprises a payment event or a non-payment event. If the number of the abnormal detection results with the payment event as the label is larger than the number of the abnormal detection results with the non-payment event as the label in the plurality of abnormal detection results, the mobile phone can determine that the label of the sample to be detected is the payment event. On the contrary, if the number of the abnormal detection results with the payment event as the label is smaller than the number of the abnormal detection results with the non-payment event as the label in the plurality of abnormal detection results, the mobile phone can determine that the label of the sample to be detected is the non-payment event.

S240, if the label of the sample to be detected is a payment event, the mobile phone pushes a payment service card.

It will be appreciated that if the sample under test is tagged as a payment event, this indicates that the user may be about to have a payment operation. In order to simplify the payment operation of the user, the mobile phone can push a payment service card. The payment service card may be used to provide a quick payment route for the user, such as a scan, a checkout code, and the like.

Through pushing the payment service card, the user can conveniently and quickly carry out payment operation, the user operation is simplified, and the user experience is improved.

Optionally, when a non-payment event is detected, the mobile phone may obtain a new normal sample. Non-payment events may include other events besides payment events, such as events that display a desktop (also may be referred to as a main interface, home interface), open non-payment applications, play video/music, display photos, and so forth.

In this embodiment of the application, when detecting that the desktop is displayed, the mobile phone may obtain user behavior data, location data, and/or environment data within a first preset time period as a new normal sample, where the first preset time period is a time period formed by taking a time when the desktop is displayed by the mobile phone as a starting point and taking a time when the starting point is moved forward by a preset time (e.g., half an hour) as an ending point. Or, the mobile phone can also obtain a newly added normal sample when other non-payment events are detected.

Optionally, the mobile phone may further collect a newly added normal sample according to a preset period. Specifically, the mobile phone can detect non-payment events in the current period, acquire user behavior data, position data and/or environment data in a preset time before each payment event occurs, and use the user behavior data, the position data and/or the environment data as samples of corresponding payment events, and use the samples as newly added normal samples.

The mobile phone can also update the normal sample set based on the newly added normal samples. Alternatively, the handset can update the normal sample set periodically (e.g., daily, weekly, monthly, etc.). Therefore, the normal samples included in the normal sample set can be fitted with the actual behaviors of the user, and the abnormality detection process is more personalized.

In summary, compared with a supervised anomaly detection mode that a large number of payment events need to be accumulated in the prior art, the method and the device for detecting the non-payment events consider that only a small number of samples of the payment events and a large number of samples of the non-payment events can be accumulated in a short time, adopt an unsupervised anomaly detection mode, regard the payment events as abnormal samples and regard the non-payment events as normal samples, can shorten data accumulation time, and can quickly realize personalized detection on different users.

An embodiment of the present application further provides a chip system, as shown in fig. 5, where the chip system includes at least one processor 501 and at least one interface circuit 502. The processor 501 and the interface circuit 502 may be interconnected by wires. For example, the interface circuit 502 may be used to receive signals from other devices (e.g., a memory of an electronic device). Also for example, interface circuit 502 may be used to send signals to other devices (e.g., processor 501).

For example, interface circuit 502 may read instructions stored in a memory in the electronic device and send the instructions to processor 501. The instructions, when executed by the processor 501, may cause the electronic device to perform the various steps in the embodiments described above.

Of course, the chip system may further include other discrete devices, which is not specifically limited in this embodiment of the present application.

Embodiments of the present application further provide a computer-readable storage medium, which includes computer instructions, and when the computer instructions are executed on an electronic device (e.g., the electronic device 200 shown in fig. 1), the electronic device 200 executes various functions or steps performed by the electronic device in the foregoing method embodiments.

Embodiments of the present application further provide a computer program product, which, when run on a computer, causes the computer to execute each function or step performed by the electronic device in the foregoing method embodiments.

From the above description of the embodiments, it is obvious for those skilled in the art to realize that the above function distribution can be performed by different function modules according to the requirement, that is, the internal structure of the device is divided into different function modules to perform all or part of the above described functions.

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

The units described as separate parts may or may not be physically separate, and parts displayed as units may be one physical unit or multiple physical units, that is, may be located in one place, or may be distributed in multiple different places. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

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

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present application may be essentially or partially contributed to by the prior art, or all or part of the technical solutions may be embodied in the form of a software product, where the software product is stored in a storage medium and includes several instructions to enable a device (which may be a single chip, a chip, or the like) or a processor (processor) to execute all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

The above description is only an embodiment of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions within the technical scope of the present disclosure should be covered by the 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 prediction method applied to an electronic device, the method comprising:

when the detection information triggers the abnormal detection condition, obtaining a sample to be detected, wherein the sample to be detected comprises user behavior data in a preset time period before the detection information triggers the abnormal detection condition;

obtaining a normal sample set, wherein the normal sample set comprises a plurality of normal samples, and the normal samples comprise user behavior data in a preset time period before a non-payment event is detected;

determining labels of the samples to be tested based on a normal sample set, wherein the labels comprise payment events and non-payment events;

and if the label of the sample to be detected is a payment event, pushing a payment service card.

2. The method of claim 1, wherein the determining the label of the test sample based on the normal sample set comprises:

inputting the sample to be detected and the normal samples into a plurality of different anomaly detection models to obtain a plurality of first anomaly scores of each sample, wherein the anomaly detection models are obtained by training historical event samples and corresponding labels of users in at least one terminal device, and the first anomaly scores are used for representing the difference degree of the two samples in different dimensions;

normalizing the plurality of first anomaly scores of each sample;

for each sample, carrying out weighted fusion on the multiple first abnormal scores subjected to normalization processing according to the weights of the multiple abnormal detection models to obtain a second abnormal score;

sequencing the second abnormal scores of all the samples according to the sequence from large to small to obtain P samples with the largest second abnormal scores;

and if the P samples comprise the samples to be detected, determining that the label of the samples to be detected is a payment event.

3. The method of claim 1, wherein the determining the label of the test sample based on the normal sample set comprises:

clustering the to-be-detected samples and the normal samples to obtain a plurality of sample clusters, wherein each sample cluster comprises a plurality of samples and a cluster center, and the samples comprise the to-be-detected samples and/or the normal samples;

calculating the distance of the feature vector from each sample to the corresponding cluster center;

obtaining N samples with the largest feature vector distance according to the feature vector distance arrangement sequence from large to small;

and if the N samples comprise the sample to be detected, determining that the label of the sample to be detected is a payment event.

4. The method of claim 3, wherein there are M sets of feature vectors between each of the samples and a corresponding cluster center, each set of feature vectors comprising a first feature vector for the sample and a second feature vector for the cluster center, and wherein calculating a feature vector distance for each sample to a corresponding cluster center comprises:

calculating the feature vector distance between a first feature vector and a second feature vector included in each group of feature vectors in the M groups of feature vectors to obtain M first feature vector distances;

carrying out normalization processing on the M first feature vector distances;

and according to the weight of the M groups of feature vectors, carrying out weighted fusion on the M first feature vector distances subjected to normalization processing to obtain the feature vector distances.

5. The method of claim 1, wherein the determining the label of the test sample based on the normal sample set comprises:

inputting the sample to be detected and the normal samples into a plurality of different anomaly detection models to obtain a plurality of anomaly detection results, wherein the anomaly detection models are obtained by training historical event samples and corresponding labels of users in at least one terminal device;

and if the number of the abnormal detection results of the payment events marked by the labels is larger than that of the abnormal detection results of the non-payment events marked by the labels, determining that the labels of the samples to be detected are the payment events.

6. The method of claim 1, wherein the determining the label of the test sample based on the normal sample set comprises:

and inputting the sample to be detected and the plurality of normal samples into an abnormal detection model to obtain the label of the sample to be detected.

7. The method according to any one of claims 2 or 5-6, wherein the anomaly detection model comprises an isolated forest algorithm-based anomaly detection model, a local anomaly factor algorithm-based anomaly detection model, or a clustering algorithm-based anomaly detection model.

8. The method according to any one of claims 1-6, wherein the detection information includes a current time, and wherein when the detection information triggers an abnormal detection condition, acquiring the sample to be tested comprises:

and if the current moment is matched with preset historical payment time, acquiring the sample to be detected, wherein the historical payment time is the time or time period of the historical payment event.

9. The method according to any one of claims 1-6, wherein the detection information includes user location information, and wherein when the detection information triggers an anomaly detection condition, acquiring the sample to be detected includes:

and if the user position information is in a preset geographic fence, acquiring the sample to be detected, wherein the position center of the geographic fence is the position of the user when the payment event occurs in history.

10. The method according to any one of claims 1-6, further comprising:

when a non-payment event is detected, acquiring a newly added normal sample; the newly added normal sample comprises user behavior data in a preset time period before a non-payment event is detected, and the non-payment event comprises events of displaying a desktop, starting a non-payment application, playing videos/music and displaying photos.

11. The method of claim 10, further comprising:

and updating the normal sample set based on the newly added normal sample according to a preset period.

12. The method of any one of claims 1-6, wherein the user behavior data comprises data reflecting whether the user is walking, running, or riding in a vehicle, whether an elevator is riding, whether WIFI is connected or disconnected, a time at which WIFI is connected or disconnected, a start time and an end time of walking, running, or riding in a vehicle, and a start time and an end time of riding in an elevator.

13. The method of any one of claims 1-6, wherein the sample further comprises location data and environmental data.

14. An electronic device comprising a processor coupled to a memory, the memory storing program instructions that, when executed by the processor, cause the electronic device to implement the method of any of claims 1-13.

15. A computer-readable storage medium comprising computer instructions;

the computer instructions, when executed on an electronic device, cause the electronic device to perform the method of any of claims 1-13.