WO2020224299A1 - 支付渠道推荐方法、装置及设备 - Google Patents

支付渠道推荐方法、装置及设备 Download PDF

Info

Publication number
WO2020224299A1
WO2020224299A1 PCT/CN2020/071867 CN2020071867W WO2020224299A1 WO 2020224299 A1 WO2020224299 A1 WO 2020224299A1 CN 2020071867 W CN2020071867 W CN 2020071867W WO 2020224299 A1 WO2020224299 A1 WO 2020224299A1
Authority
WO
WIPO (PCT)
Prior art keywords
payment
channel
target
user
recommendation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2020/071867
Other languages
English (en)
French (fr)
Inventor
徐润宇
王哲
谢仁强
苏建安
刘晖
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Advanced New Technologies Co Ltd
Original Assignee
Advanced New Technologies Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Advanced New Technologies Co Ltd filed Critical Advanced New Technologies Co Ltd
Priority to US16/803,007 priority Critical patent/US11282052B2/en
Publication of WO2020224299A1 publication Critical patent/WO2020224299A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q20/00—Payment architectures, schemes or protocols
    • G06Q20/22—Payment schemes or models
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q20/00—Payment architectures, schemes or protocols
    • G06Q20/38—Payment protocols; Details thereof
    • G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
    • G06Q20/405—Establishing or using transaction specific rules

Definitions

  • This manual relates to the field of computer technology, in particular to a method, device and equipment for recommending payment channels.
  • payment platforms often provide users with payment channels for transaction settlement, such as payment channels that users have opened in the payment platform. , Such as account balance, bank card account, credit card account, third-party wallet, coupon, red envelope, points and other channels.
  • payment channels such as account balance, bank card account, credit card account, third-party wallet, coupon, red envelope, points and other channels.
  • the user can directly use the first order payment channel provided by the payment platform to pay, or select another order Pay after the payment channel.
  • the embodiments of this specification provide a payment channel recommendation method, device, and device, which are used to recommend payment channels to users when making payments, so as to improve payment efficiency and improve user payment experience.
  • the embodiment of this specification provides a payment channel recommendation method, including:
  • the payment features including cross-features between the target user and the target payment channel, and the cross-features are used to characterize the use of the target user State the historical performance of payment by the target payment channel;
  • the recommendation score corresponding to each target payment channel among the several target payment channels is determined according to the mapping relationship.
  • the mapping relationship is determined in advance by the historical data of the payment characteristics and the historical data of the payment result characteristics.
  • the payment result feature includes the result data of whether the payment channel is selected by the user and the payment is successful;
  • the embodiment of this specification also provides a payment channel recommendation device, which includes a first determination module, an acquisition module, a second determination module, and a recommendation module;
  • the first determining module is configured to determine a target user corresponding to the payment request and several target payment channels corresponding to the payment request when a payment request is received;
  • the obtaining module is configured to obtain payment characteristics according to the target user and the several target payment channels, the payment characteristics including the cross characteristics between the target user and the target payment channels, and the cross characteristics are used to characterize The historical performance of the target user's payment using the target payment channel;
  • the second determining module is configured to determine the recommendation score corresponding to each target payment channel among the plurality of target payment channels according to a mapping relationship according to the payment feature, and the mapping relationship is preliminarily determined by the historical data of the payment feature and the payment result
  • the characteristic historical data training determines that the mapping relationship is used to characterize the relationship between the payment channel and the recommendation score, and the payment result characteristic includes the result data of whether the payment channel is selected by the user and the payment is successful;
  • the recommendation module is used to recommend the target payment channel according to the recommendation score.
  • the embodiment of this specification also provides an electronic device for recommending payment channels, including:
  • At least one processor and,
  • a memory communicatively connected with the at least one processor; wherein,
  • the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can:
  • the payment features including cross-features between the target user and the target payment channel, and the cross-features are used to characterize the use of the target user State the historical performance of payment by the target payment channel;
  • the recommendation score corresponding to each target payment channel among the several target payment channels is determined according to the mapping relationship.
  • the mapping relationship is determined in advance by the historical data of the payment characteristics and the historical data of the payment result characteristics.
  • the payment result feature includes the result data of whether the payment channel is selected by the user and the payment is successful;
  • the above-mentioned at least one technical solution adopted in the embodiment of this specification can achieve the following beneficial effects: through the mapping relationship, for each specific user in the current transaction, the payment channel is scored and recommended, because the mapping relationship is preliminarily determined by payment characteristics and payment results.
  • the historical data of the characteristics are jointly trained and determined, and the payment characteristics include the payment performance of the user using the payment channel, and the payment result characteristics include the payment success rate and the user willing rate of the payment channel.
  • the payment characteristics and the payment result characteristics are jointly trained to map the relationship, so
  • the mapping relationship can well predict the payment behavior of users using payment channels, so that it can promptly and accurately recommend payment channels that have a high payment success rate and may be selected by users, reduce the possibility of users reselecting, and improve payment efficiency , To improve the user's payment experience.
  • FIG. 1 is a schematic diagram of the overall idea of a recommendation ranking strategy of a payment channel recommendation method provided by an embodiment of this specification.
  • Figure 2 is a schematic diagram of an application environment of a payment channel recommendation method provided by an embodiment of the specification.
  • Fig. 3 is a flowchart of a method for recommending payment channels according to an embodiment of the specification.
  • Fig. 4 is a schematic structural diagram of a deep neural network in a payment channel recommendation method provided by an embodiment of the specification.
  • Fig. 5 is a schematic diagram of the system structure of a payment channel recommendation method provided by an embodiment of the specification.
  • Fig. 6 is a schematic structural diagram of a payment channel recommendation device provided by an embodiment of the specification.
  • Existing payment channel recommendation schemes often only recommend based on some payment characteristics of the payment channel, such as the payment success rate of the payment channel, the user’s preference for the payment channel, the user’s order of setting the payment channel, etc.
  • a user makes a payment, he recommends pre-ordered payment channels to the user.
  • the priority of these payment channels is preset and fixed, such as specifying the first-order payment channel, or sorting out multiple payment channels.
  • Payment channels allow users to choose. In actual payment, users often do not pay in the order of the recommended payment channels, or users choose from the recommended payment channels again, or are unwilling to choose these recommended payment channels at all , But re-select from other payment channels not recommended. In addition, users are usually unfamiliar with the transaction environment in which they are located.
  • transaction data usually contains a large amount of rich information, such as rich historical information of payment channels, rich historical information of users, and both Cross-information between information.
  • the historical information of the payment channel may include various characteristics of the payment channel used in historical payments, such as the characteristics of the user group of the payment channel, information on the payment success rate of the payment channel in different transaction environments, and the proportion of all transactions.
  • User history information can include all aspects of the user’s performance in historical payments, such as the user’s own characteristics (such as age, gender, occupation, habits, etc.), average payment amount, total transaction amount, number of transactions, type and distribution of equipment used Circumstances, whether the previous payment was successful, what payment channel was used for payment, the payment channels that have been opened, etc.; cross-information can include the historical performance of the user using the payment channel, such as the number of successful payments by the user using the payment channel, and the user using the payment channel to pay Failure time, last time the user used the payment channel to pay, whether the user used the payment channel last time, etc.
  • the inventor used the payment channel ranking strategy shown in Figure 1 in the payment channel recommendation scheme when determining the relationship between the payment channel and the recommendation score.
  • the overall idea of the ranking strategy is: extract the payment from the transaction information
  • the channel s payment success rate and user selection in historical recommendation (ie user willingness rate) are used to determine whether the historically recommended payment channel is selected by the user and whether the payment is successful, and then the determination result is used as the payment result feature.
  • Positive feedback is used to train the ranking strategy of payment channels, so that in the ranking strategy of payment channels, the input features used for training include not only payment characteristics, but also the payment success rate and user willingness rate of historically recommended payment channels.
  • payment Features can include payment performance when users use payment channels, that is, payment features include cross-features between users and payment channels, so that the sorting strategy of payment channels can be accurately and promptly paid after being trained with positive feedback Therefore, according to the target user and the target payment channel, the payment channel with a high payment success rate and which may be selected by the user is determined, and then recommended, so that the recommended output result can directly meet the business needs, which can improve the payment success rate of transaction settlement. It can also reduce the possibility of users re-selecting recommended payment channels, thereby improving payment efficiency and improving users' payment experience.
  • the terminal device 101 performs data communication with the server 102 through the network 103.
  • the network 103 may include a wide area network, a metropolitan area network, or a local area network, and other networks used for Internet communication.
  • the terminal device 101 may be used to obtain payment for transaction settlement. Requests, such as cashier counters used for transaction settlement in offline shopping malls, or computers, mobile smart phones, tablet computers and other devices used by online individuals for online shopping.
  • the server 102 can include several servers for payment. Based on the payment big data, the terminal device 101 is provided with big data information related to the payment channel.
  • the execution subject can be the terminal device 101, which is independently completed by the terminal device 101, or the terminal device 101 and the server 102, that is, the terminal device.
  • the device 101 and the server 102 are completed through the network 103 together.
  • the execution subject can be a program installed in the terminal device 101, which can be used as a payment program for transaction settlement, or can be independent Program or auxiliary program.
  • the overall process of completing payment channel recommendation independently by the terminal device 101 may be as follows:
  • the target user corresponding to the payment request and several target payment channels corresponding to the payment request are determined in this transaction payment; then, According to the target user and the plurality of target payment channels, the corresponding payment characteristics are obtained, wherein the payment characteristics include the cross characteristics between the target user and the target payment channel, and the cross characteristics are used to characterize the The performance when the target user uses the target payment channel to make payment; then, according to the payment characteristics, the recommendation score corresponding to each target payment channel among the plurality of target payment channels is determined according to the preset mapping relationship, and the mapping relationship In order to determine the correspondence between the payment channel and the recommendation score in advance based on the payment characteristics and the payment result, the payment result includes the result data of whether the payment channel is selected by the user and the payment is successful; finally, the target is recommended based on the recommendation score Payment channels to complete the settlement of this transaction.
  • mapping relationship may be provided by the server 102 to the terminal device 101 in advance, and the payment feature may also be provided by the server 102 to the terminal device 101 in advance.
  • the terminal device 101 can also follow the server 102 to update the mapping relationship and payment characteristics.
  • the overall process of completing payment channel recommendation by the terminal device 101 through the network 103 and the server 102 together may be as follows:
  • the terminal device 101 determines that the payment request is received according to the monitoring result, and sends the payment request to the server 102.
  • the server 102 determines that in this transaction payment, the target user corresponding to the payment request corresponds to the payment request And, the server 102 obtains corresponding payment characteristics according to the target user and the target payment channels, wherein the payment characteristics include the cross characteristics between the target user and the target payment channel , The cross feature is used to characterize the performance of the target user when using the target payment channel for payment; again, the server 102 determines each of the several target payment channels according to the payment feature and according to a preset mapping relationship The recommendation score corresponding to the target payment channel, the mapping relationship is the corresponding relationship between the payment channel and the recommendation score determined in advance according to the payment characteristics and the payment result, the payment result includes the result data of whether the payment channel is selected by the user and the payment is successful Finally, the server 102 recommends the target payment channel to the terminal device 101 according to the recommendation score, so as to complete the settlement of this transaction.
  • the server 102 can also preset the mapping relationship in the terminal device 101, so that the terminal device 101 can determine the recommendation score according to the mapping relationship, so that the server 102 can do big data processing such as determining payment characteristics and determining the mapping relationship. Work, I won’t repeat it here.
  • the embodiment of this specification provides a payment channel recommendation method, as shown in FIG. 3, which may include:
  • Step S102 Determine whether a payment request is received.
  • the payment request is a request for payment for transaction settlement, and may include information required for payment, such as order data, user information, and merchant information.
  • monitoring methods can be used to determine whether a payment request for transaction settlement is received according to actual application scenarios. For example, in an offline scenario, the merchant's cashier can generate a payment request based on the transaction information, and then the cashier can be monitored. Whether the station generates a payment request; for example, in an online scenario, after the user confirms to submit an order, the shopping platform generates a payment request based on the order, and then it can monitor whether the user confirms to submit the order. Therefore, through monitoring, it is possible to know in time whether a payment request is received, so that subsequent processing can be performed after the payment request is received.
  • Step S104 Determine the target user corresponding to the payment request and several target payment channels corresponding to the payment request.
  • the target user is the user who pays;
  • the target payment channel is the payment channel that may be used by the target user in the payment;
  • the payment channel can include accounts (such as account balance, account change wallet), bank cards (such as debit cards, credit cards), Coupons (such as discount coupons, redemption coupons), red envelopes, points and other payment channels.
  • the target payment channel may include the payment channel that the target user has opened in the payment platform, and it may also include the payment channel that the target user has not opened but may be willing to use.
  • the payment channels supported by the merchant when receiving payments on the payment platform and the payment channels that the user has opened in the payment platform can be determined first, and then the payment channels supported by the merchant can be set The intersection of channels and payment channels that users have opened is used as several target payment channels.
  • the payment channels that users have opened on the Alipay platform include: account balance, Huabei, Yu'ebao, bank cards and coupons.
  • the payment channels supported by merchants include: account balance, Yu'ebao, bank cards and points.
  • the account balance, Yu'ebao and bank card can be confirmed as the target payment channels corresponding to the payment request.
  • the payment channels that need to be promoted can be added to the set of target payment channels according to the characteristics of the target users, so as to promote new ones that may be willing to choose from the target users.
  • Payment channels such as promotional coupon channels based on marketing needs, can promote new payment channels and provide users with a new payment experience.
  • Step S106 Obtain payment features according to the target user and the number of target payment channels.
  • the payment feature includes the cross feature between the user and the payment channel, where the cross feature is used to characterize the historical performance of the user using the payment channel to pay, such as the number of successful payments by the user using the payment channel, and the user using the payment channel to pay.
  • the historical performance of the failure time, the last time the user used the payment channel to pay, and the last time the user used the payment channel was successful, so that these historical performances can be used as the basis for predicting the target payment channel that the target user may use in the current transaction.
  • the payment feature may also include the user feature corresponding to the user and the channel feature corresponding to the payment channel. That is to say, the user feature corresponding to the target user can be determined according to the target user, and the target can be determined according to the target payment channel. Corresponding channel characteristics of the payment channel, and determine the corresponding cross-characteristics according to the target user and the target payment channel.
  • user characteristics are used to characterize the user's historical performance in payment, such as the user's own characteristics (such as age, gender, occupation, consumption habits), average payment amount, total transaction amount, number of transactions, type of equipment used, and distribution , Whether the previous payment was successful, what payment channel was used for payment, the payment channel that has been opened, etc.; channel characteristics are used to characterize the historical performance of the payment channel used in payment, such as the user group characteristics of the payment channel, and the payment channels are different Payment success rate information in the transaction environment, information on the proportion of all transactions, etc. In this way, user characteristics and channel characteristics can also be used as payment characteristics to better characterize the historical performance of users and payment channels in payment.
  • the payment characteristics determined according to the target users and target payment channels can better reflect the possible performance of users and payment channels in the current transaction environment, and can improve the payment success rate and user willingness rate of the recommended payment channels , And can further use the payment success rate and user willingness rate for positive feedback to counter the training payment channel elimination strategy and strengthen the positive feedback effect of the ranking strategy.
  • the payment characteristics are determined from the collected transaction information, and can be selected according to the actual application target needs. For example, the application has become more mature, users and payment channels have stabilized, and more accurate information is needed. Recommend payment channels to ensure payment efficiency and user experience. Therefore, payment features can include user features, channel features, and cross-features, making payment features richer, and then recommending payment channels more accurately based on rich payment feature big data; another example At the beginning of the application, there was little user data and payment channel data, but some payment channels need to be promoted to meet user choices.
  • the payment features can include channel features and cross-features, which can reduce the payment success rate to many Recommend the payment channel selected by the user, or recommend the target payment channel to promote, to further strengthen the positive feedback of these payment channels, and finally achieve the application goal; for example, the application needs to target certain user groups, then the payment characteristics can be Including user characteristics and cross characteristics, so that the recommendation is more targeted. I will not list them all here.
  • the payment characteristics can be obtained from a database.
  • the transaction data can be processed by big data to obtain payment features from the database.
  • a log system can be used to collect transaction data, and then a big data processing platform can be used to extract payment features from the transaction data, and finally the payment features are stored in a database for query use.
  • the log system can use Logtail (Alibaba Cloud log collection agent), and the data processing platform can use ODPS (Open Data Processing Service, a distributed massive data processing platform); also, according to actual application needs, MySQL, Hbase, etc. can be used Database storage payment features, among which Hbase is a highly reliable, high-performance, column-oriented, and scalable distributed storage system. In this way, by using Hbase, a large-scale structured storage cluster can be built on a cheap PC Server (personal server) , MySQL is a relational database management system, which can increase query speed and flexibility by storing data in different tables.
  • Step S108 Determine the recommendation score corresponding to each target payment channel among the plurality of target payment channels according to the preset mapping relationship according to the payment characteristics.
  • mapping relationship (that is, the aforementioned ranking strategy) is determined in advance based on the historical data of the payment characteristics and the historical data of the payment result characteristics, and the mapping relationship is used to characterize the relationship between the payment channel and the recommendation score, and the payment result characteristics Including the result data of whether the payment channel is selected by the user and the payment is successful.
  • the payment success rate and user willingness rate of historically recommended payment channels can also be obtained from the collected historical transaction data, and the payment success rate and user willingness rate can be output to the historical output of the mapping relationship (that is, historical The recommended payment channel) to determine and generate payment result features.
  • the payment result features are used as the positive feedback input of the mapping relationship to positively feed the mapping relationship, that is, the user’s choice of each payment channel Willingness (that is, the user willingness rate) and the payment success rate (that is, the payment success rate) of each recommended payment channel at the time of final payment are jointly reflected in the mapping relationship to provide positive feedback on the recommendation process of the payment channel, so that in the transaction In the settlement payment, in the current transaction environment, the mapping relationship can be used to score each target payment channel to determine the recommendation order of each target payment channel.
  • the recommended score can be expressed in a percentile system.
  • the recommended score is 90 points, or a normalized coefficient, for example, the recommended score is 0.9, which is not specifically limited here.
  • Step S110 Recommend the target payment channel according to the recommendation score.
  • the corresponding target payment channels can be sorted and output according to the recommendation scores to respond to the payment request and facilitate the completion of transaction calculations.
  • the target payment channel corresponding to the recommendation score can be output in descending order of the recommendation score.
  • the user willingness rate and payment success rate of the historically recommended payment channels are positively fed back into the mapping relationship, so that the mapping relationship is effectively trained through positive feedback, so that when paying, according to The mapping relationship can timely and accurately determine the recommended order of each target payment channel, which can increase the success rate of payment and reduce the possibility of users re-selecting, thereby improving payment efficiency and improving users' payment experience.
  • the above steps S102-110 can be used as payment decisions, so that after the payment decision is determined to receive the payment request, it will perform the availability, visibility, sort order, and order of the target payment channel in the transaction according to the payment request. Make decisions such as default selection, and then recommend payment channels that have a high payment success rate and may be selected by the target user in the current transaction environment.
  • the target user is user A, and there are 3 target payment channels, then the payment characteristics of this transaction can be determined first, and then respectively according to the mapping relationship Determine the recommendation scores corresponding to the 3 target payment channels, and finally make recommendations based on the recommendation scores.
  • the user characteristics of user A, the channel characteristics of payment channel 1, and the intersection characteristics of user A and payment channel 1 form the first set of payment characteristics, and then use the first set of payment characteristics, according to the existing mapping
  • the relationship is to make a prediction score for payment channel 1 to obtain the recommendation score corresponding to payment channel 1.
  • the recommendation score of payment channel 1 is 90 points; similarly, according to the user characteristics of user A, the channel characteristics of payment channel 2, and user A
  • the cross feature with payment channel 2 constitutes the second set of payment feature values, and then use the second set of payment feature values to score payment channel 2 according to the existing mapping relationship.
  • the recommended score of payment channel 2 is 95 points ; And, according to the user characteristics of user A, the channel characteristics of payment channel 3 and the cross characteristics of user A and payment channel 3, constitute the third set of payment feature values, and then use the third set of payment feature values, according to the existing mapping relationship Make prediction scores for payment channel 3.
  • the recommended score for payment channel 3 is 86 points.
  • the output can be sorted according to the recommendation score (such as descending order), that is, when user A pays in this transaction, the order of the recommendation results obtained by payment channels 1 to 3 is: payment channel 2, payment channel 1, and payment channel 3. .
  • the payment success rate and user The willingness rate is at a very low level (such as zero), and the payment success rate and user willingness rate of payment channel 3 are at a high level (such as 1), so that the final payment result can be used to timely pay according to the transaction
  • the result feature is to adjust the mapping relationship, that is, the payment success rate and user willing rate corresponding to the payment channels 1 to 3, to combat the training mapping relationship.
  • the mapping relationship can more accurately reflect the relationship between the payment channel and the recommendation score, so that in the subsequent transaction payment, if the payment feature of the transaction payment is similar to the payment feature of the historical transaction (such as the same or When the similarity reaches the threshold), the similar payment channels can be recommended first, for example, payment channel 3 is also recommended.
  • the mapping relationship can be preferably obtained by a deep neural network (DNN), where the structure diagram of the DNN can be as shown in Figure 4, where a decision layer (ie, a decision network) is added to the traditional DNN, such as The hidden layer adds a judgment network, so that from the transaction data, the payment success rate and user willingness rate corresponding to the DNN output result (that is, the payment channel recommended according to the ranking strategy) can be obtained, and then the obtained payment success rate and users can be used The willingness rate, the payment result characteristics are obtained through the judgment network, and finally the DNN is jointly trained with the payment characteristics and the payment result characteristics, which not only effectively utilizes the transaction result data, but also improves the prediction accuracy of the DNN.
  • DNN deep neural network
  • the DNN output layer can be trained against the DNN output layer using the payment result characteristics as the positive feedback input characteristics, so that the output layer not only uses the historical payment characteristics for training, but also uses the historical payment result characteristics (that is, payment The success rate and user willing rate are used to train the output result judgment result), and use the current recommended payment result characteristics for confrontation training, so that after continuous confrontation training, the output result of the DNN can directly reach the application target, that is, payment Payment channels that have a high success rate and users are willing to choose can make timely and accurate decision output through the DNN output layer.
  • user characteristics, channel characteristics, and cross-features can be selected from transaction data through feature engineering according to actual application needs, and then the extracted features can be input into the connection layer (such as the hidden layer).
  • user characteristics are selected from user characteristics (such as gender, age, occupation, consumption habits, etc.), payment amount (such as each amount, total amount, etc.), number of payments, and payment equipment (such as equipment) Type, distribution of device types, etc.), payment channels (such as opened payment channels, commonly used payment channels, channels with a high payment success rate, payment channels last used, etc.); channel characteristics select the user group used by the payment channel Features (such as age, consumption habits, occupation, etc.), success rate (such as the number of successful transactions, the number of failed transactions, the time of successful payment, the time of failed payment, whether the last payment was successful, etc.), the percentage (such as all transactions)
  • the cross-characteristics select the number of successful transactions, time of success, time of failure, and recent performance (such as success or failure rate, payment amount, etc.
  • the AB test (ie AB Test) can be used to diversify the payment request for transaction settlement, that is, the AB test can be used to recommend payment channels, and the AB test can be used to perform diverging processing. It can reduce the processing requirements of the payment channel recommendation on the equipment, and can further improve the pertinence of the payment channel recommendation by collecting the transaction data after AB grouping and the transaction data to better meet the marketing needs of the actual application.
  • the recommendation score can be adjusted according to the application needs and then the recommendation can be made.
  • the recommendation score determined according to the mapping relationship can be adjusted according to the preset score adjustment strategy.
  • recommend payment channels For example, when the application needs to market to a certain type of user, it may lack user data in the early stage and the marketing effect is not satisfactory.
  • the AB test can be used to triage to obtain more data of this type of user, but the user data of this type can be obtained at this time
  • the recommendation score of the payment channel is often not high.
  • the recommendation score corresponding to the payment channel of this type of user can be adjusted. For example, in the promotion of Huabei, users who are willing to use Huabei are often young people.
  • Huabei is 60 points
  • Yu'ebao is 90 points
  • bank cards are 80 points.
  • Huabei is 60 points
  • Yu'ebao is 90 points
  • bank cards are 80 points.
  • Huabei is 60 points
  • Huabei is 60 points
  • Yu'ebao is 90 points
  • bank cards are 80 points.
  • Huabei is 60 points
  • Huabei is 90 points
  • bank cards are 80 points.
  • you can adjust Huabei’s recommendation score for example, to 85 points. This will not affect the user’s possibility of using Yu'e Bao for payment, but also Potentially used Huabei are placed in the top recommendation order to increase Huabei's exposure and provide users with a new payment option.
  • the cashier when the transaction is settled, the cashier generates a payment request.
  • the payment decision determines the target user, target payment channel, and payment characteristics, such as requesting this time from a database (such as MySQL, Hbase)
  • the target user, target payment channel and payment characteristics corresponding to the transaction, and according to the payment characteristics request the prediction result of the recommendation score from the prediction platform (such as online scoring platform), so that the payment decision can recommend the target payment to the cashier after obtaining the recommendation score Channel, so that the cashier can render the target payment channel and complete the transaction payment.
  • a log system (such as Logtail) can also be used to collect transaction-related information, such as collecting transaction information at the cash register, collecting decision results of payment decisions (such as recommendation scores, AB test diversion results, etc.), and then related to the collected transactions
  • the information is sent to a big data processing platform (such as ODPS) for big data processing, where big data processing can include feature engineering processing, transaction information association processing, etc., so that payment features and payment result features can be obtained through feature engineering processing, and features Input to the training platform of the prediction model, where the payment features and payment result features are used to train the prediction model offline (i.e. the mapping relationship).
  • the payment features can also be stored in the database for easy query and acquisition; the prediction model can also be When the application needs, provide the prediction model to the prediction platform to update the prediction model used by the prediction platform; in addition, according to the application needs, the AB test platform can be used to push the diversion strategy to the payment decision, so that the payment decision can be based on the diversion strategy. Perform AB test processing on payment request.
  • it can be divided into layers as shown in the figure (the two thick black dotted lines in the figure are layered), the AB test platform, prediction platform, and database can be set as background online processing services, and the log system payment decision can be set as foreground services , Set the big data processing platform, the prediction model training platform, and the provision of prediction models as background offline processing services.
  • the embodiments of this specification also provide a payment channel recommendation device, electronic equipment and non-volatile computer storage medium for payment channel recommendation.
  • the payment channel recommendation device 10 includes a first determination module 11, an acquisition module 12, a second determination module 13 and a recommendation module 14.
  • the first determining module 11 is used to determine the target user corresponding to the payment request and a number of target payment channels corresponding to the payment request when a payment request is received
  • the obtaining module 12 is used to determine the target user and the target payment channels corresponding to the payment request
  • a number of target payment channels acquire payment features, the payment features including cross-features between the target user and the target payment channel, and the cross-features are used to characterize the history of the target user using the target payment channel to make payments Performance
  • the second determination module 13 is used to determine the recommendation score corresponding to each target payment channel in the plurality of target payment channels according to the mapping relationship according to the payment characteristics, the mapping relationship is determined in advance by the payment characteristics and the payment result characteristic training, It is used to characterize the relationship between the payment channel and the recommendation score.
  • the payment result feature includes the result data of whether the payment channel is selected
  • the payment feature also includes user features, and the user features are used to characterize the user's historical performance in payment; at this time, the acquisition module 12 is used to: obtain the corresponding target user according to the target user The user characteristics of; and, according to the target user and the plurality of target payment channels, the cross characteristics are obtained.
  • the payment characteristics also include channel characteristics, which are used to characterize the historical performance of the payment channel used in payment; at this time, the acquisition module 12 is used to: acquire according to the several target payment channels The channel characteristic corresponding to the target payment channel; and, according to the target user and the plurality of target payment channels, the cross characteristic is obtained.
  • the payment channel recommendation device 10 may further include a collection module 15, an extraction module 16 and a determination module 17.
  • the collection module 15 is used to collect transaction information after the recommendation module 14 recommends the target payment channel according to the recommendation score; the extraction module 16 is used to extract the user corresponding to the recommended target payment channel from the transaction information The willingness rate and the payment success rate; the determination module 17 is used to generate the payment result characteristics according to the user willingness rate and the payment success rate.
  • the payment channel recommendation device 10 may further include a database module 18.
  • the extraction module 16 is also used to extract the payment feature from the transaction information; the database module 18 is used to store the payment feature in a database.
  • the payment channel recommendation device 10 may further include an AB shunt module 19.
  • the AB distribution module 19 is used to perform AB test on the payment request according to a preset distribution strategy.
  • the mapping relationship is a mapping relationship obtained according to a deep neural network.
  • the payment channel recommendation device 10 may further include a confrontation training module 20.
  • the confrontation training module 20 is configured to conduct confrontation training on the deep neural network according to the payment result characteristics of the recommended target payment channel.
  • the electronic device for recommending payment channels includes:
  • At least one processor and,
  • a memory communicatively connected with the at least one processor; wherein,
  • the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can:
  • the payment features including cross-features between the target user and the target payment channel, and the cross-features are used to characterize the use of the target user State the historical performance of payment by the target payment channel;
  • the recommendation score corresponding to each of the several target payment channels is determined according to the mapping relationship, and the mapping relationship is determined in advance by the historical data of the payment characteristics and the historical data of the payment result characteristics.
  • the mapping relationship is used to characterize the relationship between the payment channel and the recommendation score, and the payment result feature includes the result data of whether the payment channel is selected by the user and the payment is successful;
  • the embodiment of this specification provides a non-volatile computer storage medium for recommending payment channels, storing computer executable instructions, and the computer executable instructions are set as:
  • the payment characteristics including cross characteristics between the target user and the target payment channel, and the cross characteristics are used to characterize the use of the target user State the historical performance of payment by the target payment channel;
  • the recommendation score corresponding to each of the several target payment channels is determined according to the mapping relationship, and the mapping relationship is determined in advance by the historical data of the payment characteristics and the historical data of the payment result characteristics.
  • the mapping relationship is used to characterize the relationship between the payment channel and the recommendation score, and the payment result feature includes the result data of whether the payment channel is selected by the user and the payment is successful;
  • the apparatus, equipment, non-volatile computer storage medium, and method provided in the embodiments of this specification correspond to each other. Therefore, the apparatus, equipment, and non-volatile computer storage medium also have beneficial technical effects similar to the corresponding method.
  • the beneficial technical effects of the method are described in detail, therefore, the beneficial technical effects of the corresponding device, equipment, and non-volatile computer storage medium are not repeated here.
  • a programmable logic device Programmable Logic Device, PLD
  • FPGA Field Programmable Gate Array
  • HDL Hardware Description Language
  • ABEL Advanced Boolean Expression Language
  • AHDL Altera Hardware Description Language
  • HDCal JHDL
  • Lava Lava
  • Lola MyHDL
  • PALASM RHDL
  • VHDL Very-High-Speed Integrated Circuit Hardware Description Language
  • Verilog Verilog
  • the controller can be implemented in any suitable manner.
  • the controller can take the form of, for example, a microprocessor or a processor and a computer-readable medium storing computer-readable program codes (such as software or firmware) executable by the (micro)processor. , Logic gates, switches, application specific integrated circuits (ASICs), programmable logic controllers and embedded microcontrollers.
  • controllers include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicon Labs C8051F320, the memory controller can also be implemented as a part of the memory control logic.
  • controller in addition to implementing the controller in a purely computer-readable program code manner, it is entirely possible to program the method steps to make the controller use logic gates, switches, application specific integrated circuits, programmable logic controllers and embedded The same function can be realized in the form of a microcontroller, etc. Therefore, such a controller can be regarded as a hardware component, and the devices included in it for realizing various functions can also be regarded as a structure within the hardware component. Or even, the device for realizing various functions can be regarded as both a software module for realizing the method and a structure within a hardware component.
  • a typical implementation device is a computer.
  • the computer may be, for example, a personal computer, a laptop computer, a cell phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or Any combination of these devices.
  • the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.
  • a computer-usable storage media including but not limited to disk storage, CD-ROM, optical storage, etc.
  • These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing equipment to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction device.
  • the device implements the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.
  • These computer program instructions can also be loaded on a computer or other programmable data processing equipment, so that a series of operation steps are executed on the computer or other programmable equipment to produce computer-implemented processing, thereby executing on the computer or other programmable equipment
  • the instructions provide steps for implementing functions specified in a flow or multiple flows in the flowchart and/or a block or multiple blocks in the block diagram.
  • the computing device includes one or more processors (CPU), input/output interfaces, network interfaces, and memory.
  • processors CPU
  • input/output interfaces network interfaces
  • memory volatile and non-volatile memory
  • the memory may include non-permanent memory in computer readable media, random access memory (RAM) and/or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM). Memory is an example of computer readable media.
  • RAM random access memory
  • ROM read-only memory
  • flash RAM flash memory
  • Computer-readable media include permanent and non-permanent, removable and non-removable media, and information storage can be realized by any method or technology.
  • the information can be computer-readable instructions, data structures, program modules, or other data.
  • Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disc (DVD) or other optical storage, Magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media can be used to store information that can be accessed by computing devices. According to the definition in this article, computer-readable media does not include transitory media, such as modulated data signals and carrier waves.
  • program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types.
  • This application can also be practiced in distributed computing environments. In these distributed computing environments, remote processing devices connected through a communication network perform tasks.
  • program modules can be located in local and remote computer storage media including storage devices.

Landscapes

  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Finance (AREA)
  • Computer Security & Cryptography (AREA)
  • Development Economics (AREA)
  • Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种支付渠道推荐方法、装置及设备。方案包括：在接收到支付请求时，确定本次交易中的目标用户、目标支付渠道和支付特征，然后按预设的映射关系，确定本次交易中各个目标支付渠道的推荐分数，然后根据推荐分数推荐目标支付渠道，其中映射关系用于表征支付渠道与推荐分数之间的关系，预先由支付特征的历史数据和支付结果特征的历史数据训练确定，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据。

Description

支付渠道推荐方法、装置及设备 技术领域
本说明书涉及计算机技术领域，尤其涉及一种支付渠道推荐方法、装置及设备。
背景技术
当前，人们越来越频繁地使用各种支付平台，来进行交易结算的支付，这时支付平台往往会向用户提供用于交易结算的支付渠道，比如用户已在支付平台中开通过的支付渠道，如账户余额、银行卡账户、信用卡账户、第三方钱包、优惠券、红包、积分等渠道，此时，用户可以直接使用支付平台提供的第一顺序的支付渠道进行支付，或者重新选择其他顺序的支付渠道后再进行支付。
发明内容
有鉴于此，本说明书实施例提供了一种支付渠道推荐方法、装置及设备，用于在支付时，向用户推荐支付渠道，来提高支付效率和改善用户的支付体验。
本说明书实施例采用下述技术方案：
本说明书实施例提供一种支付渠道推荐方法，包括：
在接收到支付请求时，确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；
根据所述目标用户和所述若干目标支付渠道获取支付特征，所述支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付的历史表现；
根据所述支付特征，按映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系预先由所述支付特征的历史数据和支付结果特征的历史数据训练确定，用于表征支付渠道与推荐分数之间的关系，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据；
根据所述推荐分数推荐所述目标支付渠道。
本说明书实施例还提供一种支付渠道推荐装置，包括第一确定模块、获取模块、第 二确定模块和推荐模块；
所述第一确定模块用于在接收到支付请求时，确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；
所述获取模块用于根据所述目标用户和所述若干目标支付渠道获取支付特征，所述支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付的历史表现；
所述第二确定模块用于根据所述支付特征，按映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系预先由所述支付特征的历史数据和支付结果特征的历史数据训练确定，所述映射关系用于表征支付渠道与推荐分数之间的关系，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据；
所述推荐模块用于根据所述推荐分数推荐所述目标支付渠道。
本说明书实施例还提供一种用于推荐支付渠道的电子设备，包括：
至少一个处理器；以及，
与所述至少一个处理器通信连接的存储器；其中，
所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够：
在接收到支付请求时，确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；
根据所述目标用户和所述若干目标支付渠道获取支付特征，所述支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付的历史表现；
根据所述支付特征，按映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系预先由所述支付特征的历史数据和支付结果特征的历史数据训练确定，用于表征支付渠道与推荐分数之间的关系，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据；
根据所述推荐分数推荐所述目标支付渠道。
本说明书实施例采用的上述至少一个技术方案能够达到以下有益效果：通过映射关系，针对当前交易中的每个特定用户，对支付渠道进行打分后推荐，由于映射关系是预 先由支付特征和支付结果特征的历史数据共同训练确定，而支付特征包括用户使用支付渠道的支付表现，支付结果特征包括支付渠道的支付成功率和用户意愿率，这样通过将支付特征和支付结果特征共同训练映射关系，使得映射关系能很好地预测用户使用支付渠道的支付行为，从而能及时、准确地将支付成功率高且又可能被用户选择的支付渠道进行推荐，减少用户重新选择的可能，又能提高支付效率，改善用户的支付体验。
附图说明
为了更清楚地说明本说明书实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本说明书中记载的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1为本说明书实施例提供的一种支付渠道推荐方法的推荐排序策略的整体思路示意图。
图2为本说明书实施例提供的一种支付渠道推荐方法的应用环境示意图。
图3为本说明书实施例提供的一种支付渠道推荐方法的流程图。
图4为本说明书实施例提供的一种支付渠道推荐方法中深度神经网络的结构示意图。
图5为本说明书实施例提供的一种支付渠道推荐方法的系统结构示意图。
图6为本说明书实施例提供的一种支付渠道推荐装置的结构示意图。
具体实施方式
为了使本技术领域的人员更好地理解本说明书中的技术方案，下面将结合本说明书实施例中的附图，对本说明书实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本说明书实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都应当属于本申请保护的范围。
现有的支付渠道推荐方案，往往仅基于支付渠道的一些支付特性来进行推荐，比如支付特性为支付渠道的支付成功率、用户对支付渠道的喜好、用户对支付渠道的设置顺序等，这时在用户进行支付时，就向用户推荐已预先排好序的支付渠道，由于这些支付渠道的优先顺序，是预先设置好的、固定的，比如指定第一顺序的支付渠道，或者排序 出多个支付渠道让用户选择，而在实际支付中，经常有用户并未按推荐的支付渠道的顺序进行支付，或者用户重新从推荐的支付渠道中选择，又或者根本不愿意选择所推荐的这些支付渠道，而是重新从其他未推荐的支付渠道中进行选择。另外，用户对所处的交易环境通常不熟悉，在使用用户重新选择的支付渠道后，常常出现支付不畅，甚至支付失败。因此，这些支付渠道的推荐方案，推荐结果与业务目标之间存在鸿沟，也就是说，无法及时、准确地将一些支付成功率高且用户又愿意选择的渠道推荐。
基于此，发明人通过对大量的交易数据，进行了深入研究分析，发现交易数据中，通常包含有大量、丰富的信息，比如丰富的支付渠道历史信息、丰富的用户历史信息，以及这两种信息之间的交叉信息。其中，支付渠道历史信息可包括支付渠道在历史支付中被使用的各方面特征，比如支付渠道的用户群特点、支付渠道在不同交易环境中的支付成功率的信息、在全部交易中占比等；用户历史信息可包括用户在历史支付中的各方面表现，如用户自身特点(如年龄、性别、职业、习惯等)、平均支付金额、总交易金额、交易笔数、使用的设备类型及分布情况、前一次支付是否成功、使用何种支付渠道进行支付、已开通的支付渠道等；交叉信息可包括用户使用支付渠道的历史表现，如用户使用支付渠道支付成功笔数、用户使用支付渠道支付失败时间、用户使用支付渠道上次支付时间、用户使用支付渠道上次是否成功等。
因此，发明人在支付渠道的推荐方案中，在确定支付渠道与推荐分数的关系时，采用了如图1所示的支付渠道排序策略，排序策略整体思路为：从交易信息中，提取出支付渠道在历史推荐中对应的支付成功率和用户选择情况(即用户意愿率)，并用于判定历史推荐的支付渠道是否被用户选择、且是否支付成功，进而将判定结果作为支付结果特征，用于正向反馈去训练支付渠道的排序策略，这样支付渠道的排序策略中，训练所用的输入特征不仅包括了支付特性，还包括了历史推荐的支付渠道的支付成功率和用户意愿率，其中，支付特征可包括用户使用支付渠道时的支付表现，即支付特征包括了用户与支付渠道之间的交叉特征，这样支付渠道的排序策略在得到正向反馈训练后，在支付时就能够准确地、及时地，根据目标用户和目标支付渠道确定出支付成功率高且又可能被用户选择的支付渠道，然后进行推荐，从而让推荐的输出结果可以直达业务需求，既可以提高交易结算的支付成功率，又能减少用户对推荐的支付渠道进行重新选择的可能性，进而可提高支付效率，改善用户的支付体验。
本说明书实施例提供的支付渠道推荐方法的整体思路，可采用如图2所示的应用环境示意图实现。如图2所示，终端设备101通过网络103与服务端102进行数据通信， 其中网络103可包括广域网、城域网或局域网等用于互联网通信的网络，终端设备101可用于获得交易结算的支付请求，比如线下商场中用于交易结算的收银台，又比如线上个人用于网络购物的计算机、移动智能手机、平板电脑等设备，服务端102可包括若干用于支付的服务器，这样可基于支付的大数据，来向终端设备101提供支付渠道有关的大数据信息。
具体实施中，本发明实施例提供的支付渠道推荐方法，从设备角度而言，执行主体可以为终端设备101，即由终端设备101独立完成，也可以为终端设备101和服务端102，即终端设备101通过网络103与服务器102共同完成。
还有，本说明书实施例中提供的支付渠道推荐方法，从程序实现的角度而言，执行主体可以为终端设备101中安装的程序，该程序可以作为用于交易结算的支付程序，可为独立程序或者辅助程序。
一方面，由终端设备101独立完成支付渠道推荐的整体过程示意可如下：
首先，根据监测结果，确定接收到支付请求；然后，根据支付请求，去确定出本次交易支付中，所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；再然后，根据所述目标用户和所述若干目标支付渠道，去获取出对应的支付特征，其中支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付时的表现；再然后，根据所述支付特征，按预设的映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系为预先根据支付特征和支付结果确定的支付渠道与推荐分数之间的对应关系，所述支付结果包括支付渠道是否被用户选择并支付成功的结果数据；最后，根据所述推荐分数推荐所述目标支付渠道，以便完成本次交易结算。
需要说明的是，所述映射关系可由服务端102预先向终端设备101提供，所述支付特征也可以由服务端102预先向终端设备101提供。当然，在支付渠道推荐中，终端设备101还可跟随服务端102，更新所述映射关系和支付特征。
另一方面，由终端设备101通过网络103与服务端102共同来完成支付渠道推荐的整体过程示意可如下：
首先，终端设备101根据监测结果，确定接收到支付请求，并将支付请求发送给服务端102，服务端102确定出本次交易支付中，所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；以及，服务端102根据所述目标用户和所述若干目标 支付渠道，获取出对应的支付特征，其中支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付时的表现；再次，服务端102根据所述支付特征，按预设的映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系为预先根据支付特征和支付结果确定的支付渠道与推荐分数之间的对应关系，所述支付结果包括支付渠道是否被用户选择并支付成功的结果数据；最后，服务端102根据所述推荐分数向终端设备101推荐所述目标支付渠道，以便完成本次交易结算。
需要说明的是，服务端102也可将映射关系预置于终端设备101中，这样终端设备101可根据映射关系确定推荐分数，这样服务端102可做确定支付特征、确定映射关系等大数据处理工作，这里不再赘述。
以下结合附图，详细说明本申请各实施例提供的技术方案。
本说明书实施例提供一种支付渠道推荐方法，如图3所示，可包括：
步骤S102、判断是否接收到支付请求。
其中，支付请求为用于交易结算进行支付的请求，可包括有支付所需要的信息，比如订单数据、用户信息、商户信息。
具体实施中，可根据实际应用场景，采用监测方式来判断是否接收到用于交易结算的支付请求，比如在线下场景中，可由商户的收银台根据交易信息生成支付请求，这时就可监测收银台是否生成支付请求；又比如在线上场景中，用户确定提交订单后，购物平台就根据订单生成支付请求，这时就可监测用户是否确认提交订单。因此，通过监测，可及时地获知是否接收到支付请求，以便于在接收到支付请求后，做后续处理。
步骤S104、确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道。
其中，目标用户为支付的用户；目标支付渠道为支付中可能被目标用户使用的支付渠道；支付渠道可包括账户(如账户余额、账户零钱包)、银行卡(如借记卡、信用卡)、券(如折扣券、抵扣券)、红包、积分等支付时使用的渠道。
具体实施中，目标支付渠道可包括目标用户在支付平台中已开通的支付渠道，也可包括目标用户虽然未开通过、但可能愿意选用的支付渠道。
在一个实施方式中，在确定若干目标支付渠道时，可先确定出商户在支付平台收款 时所支持的支付渠道，以及用户在支付平台中已开通的支付渠道，然后可以将商户支持的支付渠道和用户已开通的支付渠道的交集，作为若干目标支付渠道。例如，用户在支付宝平台中已开通的支付渠道包括有：账户余额、花呗、余额宝、银行卡和券，商户支持的支付渠道包括有：账户余额、余额宝、银行卡和积分，这时，可以将账户余额、余额宝和银行卡确为支付请求对应的目标支付渠道。
在另一个实施方式中，在确定若干目标支付渠道时，还可根据目标用户的特点，将需要推广的支付渠道添加到目标支付渠道集合中，以便于向目标用户推广新的、可能愿意选择的支付渠道，比如根据营销需要，推广券渠道，这样既可以推广新的支付渠道，又能为用户提供新的支付体验。
步骤S106、根据所述目标用户和所述若干目标支付渠道，获取支付特征。
其中，支付特征用于表征用户和支付渠道在支付中的历史表现。具体地，支付特征包括用户与支付渠道之间的交叉特征，其中，所述交叉特征用于表征用户使用支付渠道进行支付的历史表现，比如用户使用支付渠道支付成功笔数、用户使用支付渠道支付失败时间、用户使用支付渠道上次支付时间、用户使用支付渠道上次是否成功等历史表现，这样可在当前交易中，将这些历史表现，作为预测目标用户可能使用的目标支付渠道的依据。
在一个实施方式中，所述支付特征还可包括用户对应的用户特征和支付渠道对应的渠道特征，也就是说，可根据目标用户确定出目标用户对应的用户特征，根据目标支付渠道确定出目标支付渠道对应的渠道特征，以及根据目标用户和目标支付渠道确定出对应的交叉特征。其中，用户特征用于表征用户在支付中的历史表现，如用户自身特点(如年龄、性别、职业、消费习惯)、平均支付金额、总交易金额、交易笔数、使用的设备类型及分布情况、前一次支付是否成功、使用何种支付渠道进行支付、已开通的支付渠道等；渠道特征用于表征支付渠道在支付中被使用的历史表现，比如支付渠道的用户群特点、支付渠道在不同交易环境中的支付成功率信息、在全部交易中占比信息等。这样，可将用户特征和渠道特征也作为支付特征，以更好地表征用户、支付渠道在支付中的历史表现。
通过确定出每一个用户的用户特征、每一个支付渠道的渠道特征和用户与支付渠道之间的交叉特征，这样在每一个交易中，用户有自己对应的特征值，支付渠道有自己对应的特征值，以及交易记录也会有自己对应的交叉特征值。因此，根据目标用户、目标支付渠道所确定出的支付特征，就能更好地反映出当前交易环境中的用户、支付渠道的 可能表现，可提高推荐的支付渠道的支付成功率和用户意愿率，以及可进一步利用支付成功率和用户意愿率进行正向反馈，来对抗训练支付渠道排除策略，加强排序策略的正向反馈作用。
具体实施中，支付特征从已采集的交易信息中获取确定，并可根据实际应用目标需要来进行选择，比如，应用已比较成熟，用户、支付渠道已趋于稳定，这时就需要更准确地推荐支付渠道，来保证支付效率和用户体验，因而支付特征可包括用户特征、渠道特征和交叉特征，使得支付特征更丰富，进而根据丰富的支付特征大数据，更准确地推荐支付渠道；又比如，应用刚起步时，用户数据、支付渠道数据少，但又需要大力推广一些支付渠道，以满足用户选择，这时支付特征可包括渠道特征和交叉特征，这样可将支付成功率高的、很多用户选择的支付渠道进行推荐，或者将推广的目标支付渠道进行推荐，来进一步加强这些支付渠道的正向反馈，最终达到应用目标；再比如，应用需要针对某一些用户群，这时支付特征可包括用户特征和交叉特征，这样推荐更有针对性。这里不再一一列举。
在一个实施方式中，所述支付特征可从数据库中获取。
具体实施中，可通过对交易数据进行大数据处理，来从数据库中获取支付特征。具体地，可先采用日志系统采集交易数据，再采用大数据处理平台，从交易数据中提取出支付特征，最后将支付特征存储于数据库中，以供查询使用。这里，日志系统可采用Logtail(阿里云日志采集代理)，数据处理平台可采用ODPS(Open Data Processing Service，分布式的海量数据处理平台)；还有，根据实际应用需要，可采用MySQL、Hbase等数据库存储支付特征，其中Hbase是一个高可靠性、高性能、面向列、可伸缩的分布式存储系统，这样利用Hbase，就可在廉价PC Server(个人服务器)上搭建起大规模结构化存储集群，MySQL是一种关系数据库管理系统，通过将数据保存在不同的表中，可增加查询速度和灵活性。
步骤S108、根据所述支付特征，按预设的映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数。
其中，所述映射关系(即前述的排序策略)为预先根据支付特征的历史数据和支付结果特征的历史数据确定，所述映射关系用于表征支付渠道与推荐分数之间的关系，支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据。
具体实施中，还可从采集到的历史交易数据中，获取出历史推荐的支付渠道的支付 成功率和用户意愿率，并将支付成功率和用户意愿率，对映射关系的历史输出(即历史推荐的支付渠道)进行判定，生成支付结果特征，这样将支付结果特征作为映射关系的正向反馈输入，来正向地向映射关系进行输入反馈，也就是说，将用户对各个支付渠道的选择意愿(即用户意愿率)和推荐的各个支付渠道在最终支付时的支付成功率(即支付成功率)，共同反映在映射关系中，来对支付渠道的推荐过程进行正向反馈，这样在交易结算的支付中，就可在当前交易环境中，利用所述映射关系对每个目标支付渠道进行打分，来确定每个目标支付渠道的推荐顺序。
需要说明的是，推荐分数可采用百分制表示，比如推荐分数为90分，也可采用归一化系数，比如推荐分数为0.9，这里不做具体限定。
步骤S110、根据所述推荐分数推荐所述目标支付渠道。
具体实施中，可根据推荐分数将对应的目标支付渠道进行排序输出，以响应所述支付请求，便于完成交易计算。其中，排序输出时，可依推荐分数的降序输出推荐分数对应的目标支付渠道。
在一个实施方式中，还可根据实际应用需要，推荐满足预设要求的支付渠道，来减少推荐的支付渠道的总数量，方便用户浏览和决策选择，比如，设置推荐分数阈值，然后判定所述目标支付渠道对应的所述推荐分数是否满足推荐分数阈值，进而只推荐满足推荐分数阈值的支付渠道，而未满足推荐分数阈值的支付渠道将不推荐，又比如，预设推荐数量N(N为正整数)，然后从降序排序后的支付渠道中，选取前N个进行推荐。
通过上述步骤S102～S110，通过将历史推荐的支付渠道的用户意愿率和支付成功率正向反馈到映射关系中，从而通过正向反馈对映射关系进行有效的对抗训练，这样在支付时，根据映射关系，能够及时、准确地确定出每个目标支付渠道的推荐顺序，可提高支付的成功率，减少用户重新选择的可能性，进而提高支付效率和改善用户的支付体验。
具体实施中，可将上述步骤S102～110作为支付决策，这样支付决策在确定接收到支付请求后，就根据支付请求，对本次交易中的目标支付渠道，进行可用性、可见性、排序顺序、默认选择等决策，然后将当前交易环境中，支付成功率高且又可能被目标用户选择的支付渠道进行推荐。
为便于理解，这里给出以下示例进行说明：假如在当前交易中，目标用户为用户A，目标支付渠道有3个，这时可先确定出本次交易的支付特征，然后再根据映射关系分别确定出3个目标支付渠道对应的推荐分数，最后根据推荐分数进行推荐。
具体地，就是把用户A的用户特征、支付渠道1的渠道特征、用户A与支付渠道1的交叉特征，构成第一组支付特征值，然后用第一组支付特征值，根据已有的映射关系为支付渠道1做预测打分，来获得支付渠道1对应的推荐分数，比如支付渠道1的推荐分数结果为90分；同理，根据用户A的用户特征、支付渠道2的渠道特征和用户A与支付渠道2的交叉特征，构成第二组支付特征值，然后用第二组支付特征值，根据已有的映射关系为支付渠道2做预测打分，比如支付渠道2的推荐分数结果为95分；以及，根据用户A的用户特征、支付渠道3的渠道特征和用户A与支付渠道3的交叉特征，构成第三组支付特征值，然后用第三组支付特征值，根据已有的映射关系为支付渠道3做预测打分，比如支付渠道3的推荐分数结果为86分。这样，可根据推荐分数做排序(比如降序)后输出，即用户A在本次交易支付时，所获得的支付渠道1～3的推荐结果顺序为：支付渠道2、支付渠道1和支付渠道3。
另外，在将支付渠道1～3按前述结果顺序推荐后，发现用户A最终选择了支付渠道3进行支付，且支付成功，这时对于支付渠道1和支付渠道2来说，支付成功率和用户意愿率均处于很低的水平(如为零)，而支付渠道3的支付成功率和用户意愿率处于高的水平(如为1)，这样就可利用最终支付结果，及时根据交易中的支付结果特征，去调整映射关系，即支付渠道1～3各自对应的支付成功率和用户意愿率，去对抗训练映射关系。这样在不断对抗训练后，映射关系就能更准确地反映出支付渠道与推荐分数之间的关系，从而在后续交易支付中，若交易支付的支付特征与历史交易的支付特征相近(如相同或相近程度达到阈值)时，就可优先推荐相近的支付渠道，比如也推荐支付渠道3。
在一个实施方式中，所述映射关系可优选深度神经网络(DNN)来获得，其中DNN的结构示意图可如图4所示，其中在传统的DNN中增加判定层(即判定网络)，比如在隐藏层增加判断网络，这样，在从交易数据中，获取出DNN输出结果(即根据排序策略推荐的支付渠道)对应的支付成功率和用户意愿率，然后可利用获取到的支付成功率和用户意愿率，通过判定网络获得支付结果特征，最后将支付特征和支付结果特征共同训练DNN，既有效地利用交易结果数据，又能提高DNN的预测准确性。具体地，可在输出层，利用支付结果特征作为正向反馈的输入特征对DNN输出层进行对抗训练，这样输出层不仅利用历史的支付特征进行训练，还可利用历史的支付结果特征(即支付成功率和用户意愿率对输出结果的判定结果)进行训练，以及利用当前推荐后的支付结果特征进行对抗训练，从而在不断对抗训练后，DNN的输出结果可直达应用目标，也就是说，支付成功率高且用户又愿意选择的支付渠道就能通过DNN输出层及时、准确地决策输出。
需要说明的是，用户特征、渠道特征和交叉特征可根据实际应用需要，从交易数据中通过特征工程进行选取，然后可将提取后的特征输入到连接层(如隐藏层)中。这里为便于说明和理解，用户特征选取了用户自身特点(如性别、年龄、职业、消费习惯等)、支付金额(如每笔金额、总金额等)、支付笔数、支付所用设备(如设备类型、设备类型的分布等)、支付渠道(如已开通的支付渠道、常用的支付渠道、支付成功率高的渠道、最近一次使用的支付渠道等)；渠道特征选取了支付渠道使用的用户群特点(如年龄、消费习惯、职业等)、成功率情况(如成功笔数、失败笔数、支付成功的时间、支付失败的时间、最近一次是否支付成功等)、占比率情况(如全部交易中的占比情况、成功率或失败率等)；交叉特征选取了用户使用该支付渠道进行支付的成功笔数、成功时间、失败时间、近一次表现(如是否成功、支付金额等)。
在一个优选的实施中，可采用AB测试(即AB Test)来对交易结算的支付请求进行分流处理，也就是说，可采用AB测试进行支付渠道的推荐，从而通过AB测试进行分流处理，既可降低支付渠道的推荐对设备的处理要求，又可通过采集AB分组后的交易数据，以及对交易数据进行统计，来进一步提高支付渠道推荐的针对性，更好地满足实际应用的营销需求。
具体实施中，在AB测试中确定出推荐分数后，还可根据应用需要去调整推荐分数后再进行推荐，即可按预设的分数调整策略，对根据映射关系确定出的推荐分数进行调整，然后再推荐支付渠道。比如，在应用需要针对某类用户进行营销时，早期可能缺少用户数据，营销效果不理想，这时可通过AB测试进行分流，来获取更多该类用户数据，但此时得到的该类用户的支付渠道的推荐分数，往往不高，这时就可对该类用户的支付渠道对应的推荐分数进行调整，比如在推广花呗中，愿意使用花呗的用户往往是年轻人群，这时可通过AB测试分流，将年轻用户人群导流到使用花呗的支付中，但早期时花呗对应的推荐分数可能不高，比如花呗为60分，而余额宝为90分、银行卡为80分，但营销时又想将花呗推荐给用户，这时就可将花呗的推荐分数进行调整，比如调整为85分，这样既未影响用户使用余额宝进行支付的可能，但又将用户可能使用的花呗置于靠前的推荐顺序中，来增加花呗的曝光率，还能为用户提供一种新的支付选择。
为便于理解，这里给出本说明书实施例提供的支付渠道推荐方法的一种系统结构示意说明。如图5所示，交易结算时，由收银台产生支付请求，支付决策在收到支付请求后，确定出目标用户、目标支付渠道以及支付特征，比如向数据库(如MySQL、Hbase)请求本次交易对应的目标用户、目标支付渠道及支付特征，并根据支付特征向预测平台 (如在线打分平台)请求推荐分数的预测结果，这样支付决策就能在获得推荐分数后，向收银台推荐目标支付渠道，这样收银台就能对目标支付渠道进行渲染以及完成本次交易支付。另外，还可采用日志系统(比如Logtail)，采集交易有关信息，比如采集收银台的交易信息、采集支付决策的决策结果(如推荐分数、AB测试分流结果等)，然后将采集到的交易有关信息发送给大数据处理平台(如ODPS)进行大数据处理，其中大数据处理可包括特征工程处理、交易信息的关联处理等，这样通过特征工程处理获取出支付特征和支付结果特征，还将特征输入到预测模型的训练平台中，其中支付特征和支付结果特征用于离线训练预测模型(即映射关系)，支付特征还可存储于数据库中，方便查询获取；预测模型经离线训练后，还可在应用需要时，向预测平台提供预测模型，来更新预测平台所使用的预测模型；还有，可根据应用需要，采用AB测试平台，向支付决策推送分流策略，这样支付决策可根据分流策略，对支付请求进行AB测试处理。具体实施中，可如图中分层划分(如图中的两根粗黑虚线分层)，将AB测试平台、预测平台、数据库设置为后台在线处理服务，将日志系统支付决策设置为前台服务，将大数据处理平台、预测模型训练平台、提供预测模型设置为后台离线处理服务。
基于同一个发明构思，本说明书实施例还提供用于支付渠道推荐的支付渠道推荐装置、电子设备以及非易失性计算机存储介质。
鉴于前述实施例中对所述支付渠道推荐方法已进行了详细说明，下面实施例中对装置、设备以及非易失性计算机存储介质所涉及的相应内容将不再赘述。
如图6所示，其中虚线方框表示可选的模块，所述支付渠道推荐装置10包括第一确定模块11、获取模块12、第二确定模块13和推荐模块14。其中，第一确定模块11用于在接收到支付请求时，确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；获取模块12用于根据所述目标用户和所述若干目标支付渠道获取支付特征，所述支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付的历史表现；第二确定模块13用于根据所述支付特征，按映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系预先由支付特征和支付结果特征训练确定，用于表征支付渠道与推荐分数之间的关系，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据；推荐模块14用于根据所述推荐分数推荐所述目标支付渠道。
可选地，所述支付特征还包括用户特征，所述用户特征用于表征用户在支付中的历史表现；这时，获取模块12就用于：根据所述目标用户，获取所述目标用户对应的所 述用户特征；以及，根据所述目标用户和所述若干目标支付渠道，获取所述交叉特征。
可选地，所述支付特征还包括渠道特征，所述渠道特征用于表征支付渠道在支付中被使用的历史表现；这时，获取模块12就用于：根据所述若干目标支付渠道，获取所述目标支付渠道对应的所述渠道特征；以及，根据所述目标用户和所述若干目标支付渠道，获取所述交叉特征。
可选地，支付渠道推荐装置10还可包括采集模块15、提取模块16和判定模块17。其中，采集模块15用于在推荐模块14根据所述推荐分数推荐所述目标支付渠道后，采集交易信息；提取模块16用于从所述交易信息中，提取所推荐的目标支付渠道对应的用户意愿率和支付成功率；判定模块17用于根据所述用户意愿率和所述支付成功率，生成所述支付结果特征。
可选地，支付渠道推荐装置10还可包括数据库模块18。其中，提取模块16还用于从所述交易信息中，提取所述支付特征；数据库模块18用于将所述支付特征存储于数据库中。
可选地，支付渠道推荐装置10还可包括AB分流模块19。其中，AB分流模块19用于按预设的分流策略对所述支付请求进行AB测试。
可选地，所述映射关系为根据深度神经网络得到的映射关系，这时支付渠道推荐装置10还可包括对抗训练模块20。其中，对抗训练模块20用于根据所推荐的目标支付渠道的所述支付结果特征，对所述深度神经网络进行对抗训练。
基于同一个发明构思，本说明书实施例中提供的用于推荐支付渠道的电子设备，包括：
至少一个处理器；以及，
与所述至少一个处理器通信连接的存储器；其中，
所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够：
在接收到支付请求时，确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；
根据所述目标用户和所述若干目标支付渠道获取支付特征，所述支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使 用所述目标支付渠道进行支付的历史表现；
根据所述支付特征，按映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系预先由所述支付特征的历史数据和支付结果特征的历史数据训练确定，所述映射关系用于表征支付渠道与推荐分数之间的关系，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据；
根据所述推荐分数推荐所述目标支付渠道。
基于同样的思路，本说明书实施例提供一种用于推荐支付渠道的非易失性计算机存储介质，存储有计算机可执行指令，所述计算机可执行指令设置为：
在接收到支付请求时，确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；
根据所述目标用户和所述若干目标支付渠道获取支付特征，所述支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付的历史表现；
根据所述支付特征，按映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系预先由所述支付特征的历史数据和支付结果特征的历史数据训练确定，所述映射关系用于表征支付渠道与推荐分数之间的关系，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据；
根据所述推荐分数推荐所述目标支付渠道。
上述对本说明书特定实施例进行了描述。其它实施例在所附权利要求书的范围内。在一些情况下，在权利要求书中记载的动作或步骤可以按照不同于实施例中的顺序来执行并且仍然可以实现期望的结果。另外，在附图中描绘的过程不一定要求示出的特定顺序或者连续顺序才能实现期望的结果。在某些实施方式中，多任务处理和并行处理也是可以的或者可能是有利的。
本说明书中的各个实施例均采用递进的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于装置、设备、非易失性计算机存储介质实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
本说明书实施例提供的装置、设备、非易失性计算机存储介质与方法是对应的， 因此，装置、设备、非易失性计算机存储介质也具有与对应方法类似的有益技术效果，由于上面已经对方法的有益技术效果进行了详细说明，因此，这里不再赘述对应装置、设备、非易失性计算机存储介质的有益技术效果。
在20世纪90年代，对于一个技术的改进可以很明显地区分是硬件上的改进(例如，对二极管、晶体管、开关等电路结构的改进)还是软件上的改进(对于方法流程的改进)。然而，随着技术的发展，当今的很多方法流程的改进已经可以视为硬件电路结构的直接改进。设计人员几乎都通过将改进的方法流程编程到硬件电路中来得到相应的硬件电路结构。因此，不能说一个方法流程的改进就不能用硬件实体模块来实现。例如，可编程逻辑器件(Programmable Logic Device,PLD)(例如现场可编程门阵列(Field Programmable Gate Array，FPGA))就是这样一种集成电路，其逻辑功能由用户对器件编程来确定。由设计人员自行编程来把一个数字系统“集成”在一片PLD上，而不需要请芯片制造厂商来设计和制作专用的集成电路芯片。而且，如今，取代手工地制作集成电路芯片，这种编程也多半改用“逻辑编译器(logic compiler)”软件来实现，它与程序开发撰写时所用的软件编译器相类似，而要编译之前的原始代码也得用特定的编程语言来撰写，此称之为硬件描述语言(Hardware Description Language，HDL)，而HDL也并非仅有一种，而是有许多种，如ABEL(Advanced Boolean Expression Language)、AHDL(Altera Hardware Description Language)、Confluence、CUPL(Cornell University Programming Language)、HDCal、JHDL(Java Hardware Description Language)、Lava、Lola、MyHDL、PALASM、RHDL(Ruby Hardware Description Language)等，目前最普遍使用的是VHDL(Very-High-Speed Integrated Circuit Hardware Description Language)与Verilog。本领域技术人员也应该清楚，只需要将方法流程用上述几种硬件描述语言稍作逻辑编程并编程到集成电路中，就可以很容易得到实现该逻辑方法流程的硬件电路。
控制器可以按任何适当的方式实现，例如，控制器可以采取例如微处理器或处理器以及存储可由该(微)处理器执行的计算机可读程序代码(例如软件或固件)的计算机可读介质、逻辑门、开关、专用集成电路(Application Specific Integrated Circuit，ASIC)、可编程逻辑控制器和嵌入微控制器的形式，控制器的例子包括但不限于以下微控制器：ARC 625D、Atmel AT91SAM、Microchip PIC18F26K20以及Silicone Labs C8051F320，存储器控制器还可以被实现为存储器的控制逻辑的一部分。本领域技术人员也知道，除了以纯计算机可读程序代码方式实现控制器以外，完全可以通过将方法步骤进行逻辑编程来使得控制器以逻辑门、开关、专用集成电路、可编程逻辑控制器和嵌入微控制器等的形式来实现相同功能。因此这种控制器可以被认为是一种硬件部件，而 对其内包括的用于实现各种功能的装置也可以视为硬件部件内的结构。或者甚至，可以将用于实现各种功能的装置视为既可以是实现方法的软件模块又可以是硬件部件内的结构。
上述实施例阐明的系统、装置、模块或单元，具体可以由计算机芯片或实体实现，或者由具有某种功能的产品来实现。一种典型的实现设备为计算机。具体的，计算机例如可以为个人计算机、膝上型计算机、蜂窝电话、相机电话、智能电话、个人数字助理、媒体播放器、导航设备、电子邮件设备、游戏控制台、平板计算机、可穿戴设备或者这些设备中的任何设备的组合。
为了描述的方便，描述以上装置时以功能分为各种单元分别描述。当然，在实施本申请时可以把各单元的功能在同一个或多个软件和/或硬件中实现。
本领域内的技术人员应明白，本发明的实施例可提供为方法、系统、或计算机程序产品。因此，本发明可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本发明可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本发明是参照根据本发明实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或 方框图一个方框或多个方框中指定的功能的步骤。
在一个典型的配置中，计算设备包括一个或多个处理器(CPU)、输入/输出接口、网络接口和内存。
内存可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)。内存是计算机可读介质的示例。
计算机可读介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括，但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带，磁带磁磁盘存储或其他磁性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。按照本文中的界定，计算机可读介质不包括暂存电脑可读媒体(transitory media)，如调制的数据信号和载波。
还需要说明的是，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、商品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、商品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、商品或者设备中还存在另外的相同要素。
本申请可以在由计算机执行的计算机可执行指令的一般上下文中描述，例如程序模块。一般地，程序模块包括执行特定任务或实现特定抽象数据类型的例程、程序、对象、组件、数据结构等等。也可以在分布式计算环境中实践本申请，在这些分布式计算环境中，由通过通信网络而被连接的远程处理设备来执行任务。在分布式计算环境中，程序模块可以位于包括存储设备在内的本地和远程计算机存储介质中。
本说明书中的各个实施例均采用递进的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于系统实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
以上所述仅为本申请的实施例而已，并不用于限制本申请。对于本领域技术人员来说，本申请可以有各种更改和变化。凡在本申请的精神和原理之内所作的任何修改、等同替换、改进等，均应包含在本申请的权利要求范围之内。

Claims (15)

  1. 一种支付渠道推荐方法，包括：
    在接收到支付请求时，确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；
    根据所述目标用户和所述若干目标支付渠道获取支付特征，所述支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付的历史表现；
    根据所述支付特征，按映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系预先由所述支付特征的历史数据和支付结果特征的历史数据训练确定，用于表征支付渠道与推荐分数之间的关系，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据；
    根据所述推荐分数推荐所述目标支付渠道。
  2. 如权利要求1所述的支付渠道推荐方法，所述支付特征还包括用户特征，所述用户特征用于表征用户在支付中的历史表现。
  3. 如权利要求1所述的支付渠道推荐方法，所述支付特征还包括渠道特征，所述渠道特征用于表征支付渠道在支付中被使用的历史表现。
  4. 如权利要求1所述的支付渠道推荐方法，在根据所述推荐分数推荐所述目标支付渠道后，所述支付渠道推荐方法还包括：
    采集交易信息；
    从所述交易信息中，提取所推荐的目标支付渠道对应的用户意愿率和支付成功率；
    根据所述用户意愿率和所述支付成功率，生成所述支付结果特征。
  5. 如权利要求4所述的支付渠道推荐方法，在采集交易信息后，所述支付渠道推荐方法还包括：
    从所述交易信息中，提取所述支付特征；
    将所述支付特征存储于数据库中。
  6. 如权利要求1所述的支付渠道推荐方法，所述支付渠道推荐方法还包括：按预设的分流策略对所述支付请求进行AB测试。
  7. 如权利要求1所述的支付渠道推荐方法，所述映射关系为根据深度神经网络得到的映射关系；
    所述支付渠道推荐方法还包括：根据所推荐的目标支付渠道的所述支付结果特征，对所述深度神经网络进行对抗训练。
  8. 一种支付渠道推荐装置，包括第一确定模块、获取模块、第二确定模块和推荐模块；
    所述第一确定模块用于在接收到支付请求时，确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；
    所述获取模块用于根据所述目标用户和所述若干目标支付渠道获取支付特征，所述支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付的历史表现；
    所述第二确定模块用于根据所述支付特征，按映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系预先由所述支付特征的历史数据和支付结果特征的历史数据训练确定，用于表征支付渠道与推荐分数之间的关系，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据；
    所述推荐模块用于根据所述推荐分数推荐所述目标支付渠道。
  9. 如权利要求8所述的支付渠道推荐装置，所述支付特征还包括用户特征，所述用户特征用于表征用户在支付中的历史表现。
  10. 如权利要求8所述的支付渠道推荐装置，所述支付特征还包括渠道特征，所述渠道特征用于表征支付渠道在支付中被使用的历史表现。
  11. 如权利要求8所述的支付渠道推荐装置，所述支付渠道推荐装置还包括采集模块、提取模块和判定模块；
    所述采集模块用于在所述推荐模块根据所述推荐分数推荐所述目标支付渠道后，采集交易信息；
    所述提取模块用于从所述交易信息中，提取所推荐的目标支付渠道对应的用户意愿率和支付成功率；
    所述判定模块用于根据所述用户意愿率和所述支付成功率，生成所述支付结果特征。
  12. 如权利要求11所述的支付渠道推荐装置，所述支付渠道推荐装置还包括数据库模块；
    所述提取模块还用于从所述交易信息中，提取所述支付特征；
    所述数据库模块用于将所述支付特征存储于数据库中。
  13. 如权利要求8所述的支付渠道推荐装置，所述支付渠道推荐装置还包括AB分流模块，所述AB分流模块用于按预设的分流策略对所述支付请求进行AB测试。
  14. 如权利要求8所述的支付渠道推荐装置，所述映射关系为根据深度神经网络得到的映射关系，所述支付渠道推荐装置还包括对抗训练模块，所述对抗训练模块用于根据 所推荐的目标支付渠道的所述支付结果特征，对所述深度神经网络进行对抗训练。
  15. 一种用于推荐支付渠道的电子设备，包括：
    至少一个处理器；以及，
    与所述至少一个处理器通信连接的存储器；其中，
    所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够：
    在接收到支付请求时，确定所述支付请求对应的目标用户和所述支付请求对应的若干目标支付渠道；
    根据所述目标用户和所述若干目标支付渠道获取支付特征，所述支付特征包括所述目标用户与所述目标支付渠道之间的交叉特征，所述交叉特征用于表征所述目标用户使用所述目标支付渠道进行支付的历史表现；
    根据所述支付特征，按映射关系确定所述若干目标支付渠道中各个目标支付渠道对应的推荐分数，所述映射关系预先由所述支付特征的历史数据和支付结果特征的历史数据训练确定，用于表征支付渠道与推荐分数之间的关系，所述支付结果特征包括支付渠道是否被用户选择并支付成功的结果数据；
    根据所述推荐分数推荐所述目标支付渠道。
PCT/CN2020/071867 2019-05-06 2020-01-14 支付渠道推荐方法、装置及设备 Ceased WO2020224299A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/803,007 US11282052B2 (en) 2019-05-06 2020-02-27 Payment channel recommendation

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910377788.0A CN110245935B (zh) 2019-05-06 2019-05-06 支付渠道推荐方法、装置及设备
CN201910377788.0 2019-05-06

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/803,007 Continuation US11282052B2 (en) 2019-05-06 2020-02-27 Payment channel recommendation

Publications (1)

Publication Number Publication Date
WO2020224299A1 true WO2020224299A1 (zh) 2020-11-12

Family

ID=67883760

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/071867 Ceased WO2020224299A1 (zh) 2019-05-06 2020-01-14 支付渠道推荐方法、装置及设备

Country Status (2)

Country Link
CN (1) CN110245935B (zh)
WO (1) WO2020224299A1 (zh)

Families Citing this family (27)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11282052B2 (en) * 2019-05-06 2022-03-22 Advanced New Technologies Co., Ltd. Payment channel recommendation
CN110245935B (zh) * 2019-05-06 2021-03-30 创新先进技术有限公司 支付渠道推荐方法、装置及设备
CN110689336A (zh) * 2019-09-23 2020-01-14 阿里巴巴集团控股有限公司 一种支付渠道的决策方法、装置及电子设备
CN110908746A (zh) * 2019-10-12 2020-03-24 平安银行股份有限公司 一种数据处理方法、系统、可读存储介质及终端设备
CN111091358B (zh) * 2019-12-16 2024-04-16 中国建设银行股份有限公司 多支付渠道的统一处理方法及系统
CN111144874B (zh) * 2019-12-20 2023-09-26 支付宝实验室(新加坡)有限公司 支付方式推荐方法、装置及设备
CN111178865A (zh) * 2019-12-31 2020-05-19 中国银行股份有限公司 一种支付方法及装置
CN111242633A (zh) * 2020-01-07 2020-06-05 支付宝(杭州)信息技术有限公司 信息提示方法、装置、设备及介质
CN113139842A (zh) * 2020-01-17 2021-07-20 阿里巴巴集团控股有限公司 一种表单处理方法、装置和系统
US11468415B2 (en) * 2020-03-17 2022-10-11 Bank Of America Corporation Automated transaction processing based on cognitive learning
CN111784384B (zh) * 2020-06-19 2022-07-19 支付宝(杭州)信息技术有限公司 支付业务数据处理方法、装置、设备及系统
CN111754218A (zh) * 2020-06-29 2020-10-09 深圳前海微众银行股份有限公司 支付方式推荐方法和装置
CN111882318A (zh) * 2020-07-06 2020-11-03 中国银联股份有限公司 一种基于神经网络的智能卡片推荐方法及装置
CN111932238B (zh) * 2020-07-27 2024-03-22 中国工商银行股份有限公司 支付账户推荐方法、装置及电子设备
CN112101953A (zh) * 2020-09-28 2020-12-18 中国银行股份有限公司 一种利用最优支付链路的移动支付处理方法及系统
CN112597385B (zh) * 2020-12-16 2025-08-26 北京三快在线科技有限公司 一种内容推送方法、装置、介质及电子设备
CN113159740B (zh) * 2021-03-11 2024-05-17 深圳市分期乐网络科技有限公司 一种支付通道确定方法、装置、电子设备及存储介质
CN115147105A (zh) * 2021-03-30 2022-10-04 厦门市易联众易惠科技有限公司 一种支付渠道推荐方法、装置、设备和存储介质
CN113222566B (zh) * 2021-05-07 2024-11-26 深圳乐信软件技术有限公司 一种支付管理系统、方法、装置及存储介质
CN113112263B (zh) * 2021-05-07 2025-03-11 深圳乐信软件技术有限公司 一种支付方式的展示方法、装置、服务器及存储介质
CN113435871B (zh) * 2021-06-29 2025-01-03 广东粤科芮银信息科技有限公司 支付渠道异常检查方法、装置、计算设备及存储介质
CN113935734A (zh) * 2021-09-15 2022-01-14 北京讯通安添通讯科技有限公司 终端安全支付方式管理的方法、装置、终端和存储介质
CN114548964A (zh) * 2022-02-10 2022-05-27 支付宝(杭州)信息技术有限公司 一种支付方法、装置及设备
CN114511313A (zh) * 2022-02-24 2022-05-17 平安普惠企业管理有限公司 支付通道接入方法、系统、计算机设备及可读存储介质
CN115170107A (zh) * 2022-08-02 2022-10-11 广域铭岛数字科技有限公司 支付优先级调整方法、系统、电子设备及可读存储介质
CN115170099B (zh) * 2022-09-08 2022-12-20 国网汇通金财（北京）信息科技有限公司 一种支付渠道确定方法及系统
CN116308332A (zh) * 2022-12-13 2023-06-23 阿里巴巴（中国）有限公司 支付方式推荐方法、装置及电子设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140310159A1 (en) * 2013-04-10 2014-10-16 Fair Isaac Corporation Reduced fraud customer impact through purchase propensity
CN106355391A (zh) * 2015-07-16 2017-01-25 阿里巴巴集团控股有限公司 一种业务处理方法及装置
CN107679849A (zh) * 2017-10-09 2018-02-09 马上消费金融股份有限公司 一种支付渠道选择方法、系统及一种计算机可读存储介质
CN110245935A (zh) * 2019-05-06 2019-09-17 阿里巴巴集团控股有限公司 支付渠道推荐方法、装置及设备

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107464108A (zh) * 2016-06-03 2017-12-12 上海点融信息科技有限责任公司 自动选择支付渠道的方法和设备
CN107403316A (zh) * 2017-08-03 2017-11-28 广州爱九游信息技术有限公司 筛选支付方式的方法、装置、计算设备和存储介质
CN108399195A (zh) * 2018-01-29 2018-08-14 阿里巴巴集团控股有限公司 支付渠道的推荐方法及装置
CN114418568A (zh) * 2018-04-02 2022-04-29 创新先进技术有限公司 一种支付方式推荐方法、装置及设备
CN109146451A (zh) * 2018-07-05 2019-01-04 麒麟合盛网络技术股份有限公司 一种支付渠道推荐方法和装置
CN109359961A (zh) * 2018-08-23 2019-02-19 阿里巴巴集团控股有限公司 清算渠道推荐方法及装置和电子设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140310159A1 (en) * 2013-04-10 2014-10-16 Fair Isaac Corporation Reduced fraud customer impact through purchase propensity
CN106355391A (zh) * 2015-07-16 2017-01-25 阿里巴巴集团控股有限公司 一种业务处理方法及装置
CN107679849A (zh) * 2017-10-09 2018-02-09 马上消费金融股份有限公司 一种支付渠道选择方法、系统及一种计算机可读存储介质
CN110245935A (zh) * 2019-05-06 2019-09-17 阿里巴巴集团控股有限公司 支付渠道推荐方法、装置及设备

Also Published As

Publication number Publication date
CN110245935A (zh) 2019-09-17
CN110245935B (zh) 2021-03-30

Similar Documents

Publication Publication Date Title
CN110245935B (zh) 支付渠道推荐方法、装置及设备
US11282052B2 (en) Payment channel recommendation
CN110413877B (zh) 一种资源推荐方法、装置及电子设备
US11373161B2 (en) Post-paid transaction data processing method and device, processing apparatus, and server
US10748152B2 (en) Providing history-based data processing
TWI650653B (zh) 大數據處理方法及平台
CN108460681B (zh) 一种风险管控方法及装置
TW201942826A (zh) 支付方式推薦方法、裝置及設備
CN111309614A (zh) A/b测试方法、装置及电子设备
CN110443618A (zh) 风控策略的生成方法及装置
CN113159834B (zh) 一种商品信息排序方法、装置以及设备
CN110020427A (zh) 策略确定方法和装置
CN114329207A (zh) 多业务信息排序系统、方法、存储介质及电子设备
CN116485391A (zh) 支付推荐处理方法及装置
WO2024198539A1 (zh) 支付处理方法及装置
CN111062770B (zh) 商户识别方法、设备及计算机可读介质
CN113297462A (zh) 数据处理方法、装置、设备和存储介质
CN111325557A (zh) 一种商户风险的检测方法、装置及设备
CN118396619A (zh) 推荐处理方法及装置
CN110415039A (zh) 业务处理的方法及装置
CN111798282A (zh) 一种信息处理方法、终端及存储介质
HK40014382B (zh) 支付渠道推荐方法、装置及设备
US20160275529A1 (en) Method and system for recommending a competitor
US20210256509A1 (en) Planning currency exchange transactions based on international travel budgeting information
CN111400604A (zh) 数据处理、针对目标对象的行为推荐方法、装置及设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20802003

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20802003

Country of ref document: EP

Kind code of ref document: A1