CN116523069A - Coupon pushing method and device based on federal modeling - Google Patents
Coupon pushing method and device based on federal modeling Download PDFInfo
- Publication number
- CN116523069A CN116523069A CN202310386490.2A CN202310386490A CN116523069A CN 116523069 A CN116523069 A CN 116523069A CN 202310386490 A CN202310386490 A CN 202310386490A CN 116523069 A CN116523069 A CN 116523069A
- Authority
- CN
- China
- Prior art keywords
- data
- initiator
- user
- characteristic data
- server
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000027455 binding Effects 0.000 claims abstract description 112
- 239000003999 initiator Substances 0.000 claims description 94
- 238000004590 computer program Methods 0.000 claims description 27
- 230000006870 function Effects 0.000 claims description 14
- 238000003860 storage Methods 0.000 claims description 10
- 230000006399 behavior Effects 0.000 claims description 8
- 230000004927 fusion Effects 0.000 claims description 4
- 238000009826 distribution Methods 0.000 claims description 2
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 9
- 239000000047 product Substances 0.000 description 9
- 230000015572 biosynthetic process Effects 0.000 description 7
- 238000003786 synthesis reaction Methods 0.000 description 7
- 238000004891 communication Methods 0.000 description 6
- 230000010354 integration Effects 0.000 description 6
- 230000007246 mechanism Effects 0.000 description 6
- 238000012545 processing Methods 0.000 description 5
- 238000007726 management method Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 230000008520 organization Effects 0.000 description 3
- 230000004308 accommodation Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000007115 recruitment Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 240000007594 Oryza sativa Species 0.000 description 1
- 235000007164 Oryza sativa Nutrition 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 230000003796 beauty Effects 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 235000014510 cooky Nutrition 0.000 description 1
- 239000010779 crude oil Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 239000010931 gold Substances 0.000 description 1
- 229910052737 gold Inorganic materials 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000010970 precious metal Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 235000009566 rice Nutrition 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000005406 washing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0207—Discounts or incentives, e.g. coupons or rebates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0207—Discounts or incentives, e.g. coupons or rebates
- G06Q30/0239—Online discounts or incentives
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Accounting & Taxation (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a coupon pushing method and device based on federal modeling, and relates to the technical field of artificial intelligence, wherein the method comprises the following steps: acquiring characteristic data of a user to be predicted in a financial institution and characteristic data of an external internet company; inputting the characteristic data into a binding probability prediction model to obtain the binding probability of a user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company; and pushing the coupons to the user groups which are most suitable for issuing according to the binding probability of the users to be predicted on the public numbers of the financial institutions. The invention can expand the characteristic data provided by external Internet companies, and predict the binding probability of users on the public numbers of financial institutions by combining the characteristic data of the institutions in a federal modeling mode, so as to push coupons to the user groups most suitable for issuing.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a coupon pushing method and device based on federal modeling.
Background
This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. The description herein is not admitted to be prior art by inclusion in this section.
Currently, conventional financial institutions have adopted a variety of promotional approaches to pull new customers and prevent customer chunking, with pushing coupons (coupons, merchant coupons, etc.) to users being one of the approaches taken. The traditional coupon pushing method comprises the following steps: is widely distributed for fixed people; or, the financial institution is adopted to preserve data for machine learning modeling, the issuing crowd is predicted to issue, and the existing coupon pushing mode is inaccurate.
Disclosure of Invention
The embodiment of the invention provides a coupon pushing method based on federal modeling, which is used for expanding characteristic data provided by an external internet company, predicting the binding probability of a user on a public number of a financial institution by combining the characteristic data of the organization in a federal modeling mode, and pushing coupons to a user group most suitable for issuing, and comprises the following steps:
acquiring characteristic data of a user to be predicted in a financial institution and characteristic data of an external internet company;
inputting the characteristic data of the user to be predicted in the financial institution and the characteristic data of the external internet company into a pre-established binding probability prediction model to obtain the binding probability of the user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company;
And pushing the coupons to the user groups which are most suitable for issuing according to the binding probability of the users to be predicted on the public numbers of the financial institutions.
The embodiment of the invention also provides a coupon pushing device based on federal modeling, which is used for expanding characteristic data provided by an external internet company, predicting the binding probability of a user on a public number of a financial institution by combining the characteristic data of the institution in a federal modeling mode, and pushing coupons to a user group which is most suitable for issuing, and comprises the following steps:
the acquiring unit is used for acquiring the characteristic data of the user to be predicted in the financial institution and the characteristic data of the external internet company;
the prediction unit is used for inputting the characteristic data of the user to be predicted in the financial institution and the characteristic data of the external internet company into a pre-established binding probability prediction model to obtain the binding probability of the user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company;
and the pushing unit is used for pushing the coupons to the user groups which are most suitable for being issued according to the binding probability of the users to be predicted on the public numbers of the financial institutions.
The embodiment of the invention also provides computer equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the coupon pushing method based on federal modeling when executing the computer program.
The embodiment of the invention also provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program realizes the coupon pushing method based on federal modeling when being executed by a processor.
The embodiment of the invention also provides a computer program product, which comprises a computer program, wherein the computer program realizes the coupon pushing method based on federal modeling when being executed by a processor.
In the embodiment of the invention, compared with the technical scheme of low coupon pushing precision in the prior art, the coupon pushing scheme based on federal modeling is characterized by comprising the following steps: acquiring characteristic data of a user to be predicted in a financial institution and characteristic data of an external internet company; inputting the characteristic data into a binding probability prediction model to obtain the binding probability of a user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company; according to the binding probability of the user to be predicted on the public number of the financial institution, pushing the coupon to the user group which is most suitable for being issued, realizing that the characteristic data provided by an external Internet company can be expanded, and according to the characteristic data of the mechanism, predicting the binding probability of the user on the public number of the financial institution in a federal modeling mode, so as to push the coupon to the user group which is most suitable for being issued.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. In the drawings:
FIG. 1 is a flow chart of a federal modeling-based coupon pushing method in an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a pre-established binding card probability prediction model for federal learning in an embodiment of the present invention;
FIG. 3 is a flow chart of pushing coupons to the most suitable user groups for distribution in an embodiment of the present invention;
FIG. 4 is a schematic diagram of a federally modeled coupon pushing device according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a computer device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the embodiments of the present invention will be described in further detail with reference to the accompanying drawings. The exemplary embodiments of the present invention and their descriptions herein are for the purpose of explaining the present invention, but are not to be construed as limiting the invention.
In the technical scheme, the acquisition, storage, use, processing and the like of the data all accord with the relevant regulations of laws and regulations.
According to the coupon pushing scheme based on federal modeling, which is provided by the embodiment of the invention, a scheme of pulling a new binding card on a public number by a bank customer based on federal modeling can be realized, in the scheme, in order to better describe customer figures, customers can be more accurately known and customer preferences can be captured, customer data dimension can be expanded on the premise of protecting data safety and privacy in a federal modeling mode, and customer groups which are more required by business can be more circled. The federally modeled coupon pushing scheme is described in detail below.
Fig. 1 is a flow chart of a coupon pushing method based on federal modeling according to an embodiment of the present invention, as shown in fig. 1, the method includes the following steps:
step 101: acquiring characteristic data of a user to be predicted in a financial institution and characteristic data of an external internet company;
step 102: inputting the characteristic data of the user to be predicted in the financial institution and the characteristic data of the external internet company into a pre-established binding probability prediction model to obtain the binding probability of the user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company;
Step 103: pushing the coupons to the user groups most suitable for issuing according to the binding probability of the users to be predicted on the public numbers of the financial institutions; the most suitable group of users (guest groups) may be those who prefer to unbind from the incentive of coupons, share more, and chain the activity.
The coupon pushing method based on federal modeling provided by the embodiment of the invention works as follows: acquiring characteristic data of a user to be predicted in a financial institution and characteristic data of an external internet company; inputting the characteristic data into a binding probability prediction model to obtain the binding probability of a user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company; and pushing the coupons to the user groups which are most suitable for issuing according to the binding probability of the users to be predicted on the public numbers of the financial institutions.
Compared with the technical scheme of low coupon pushing precision in the prior art, the coupon pushing method based on federal modeling provided by the embodiment of the invention can expand the feature data provided by external Internet companies, and by combining the feature data of the mechanism, the binding probability of users on the public numbers of financial institutions is predicted in a federal modeling mode, so that coupons are pushed to user groups which are most suitable for issuing. The federal modeling-based coupon pushing method is described in detail below.
The embodiment of the invention provides a coupon pushing method based on federal modeling, which is different from a commonly issued form in that the method is used for defining guest groups based on federal modeling, and the guest groups are stimulated by issuing coupons (full reduction of money), so that on the basis of the existing financial data of a traditional financial institution, the user characteristics are widened by adopting the federal mode, and the Internet data are supplemented.
By adopting federal modeling, internet data (such as high consumption, high user activity, strong social influence (like sharing) and the like of other card users are bound) can be longitudinally supplemented for federal learning, a model is built, and a customer group which is more suitable for issuing coupons is predicted. Federal modeling can supplement or optimize the following guest groups:
1. there are customers who are accustomed to paying with other bank cards but do not use the present line card.
2. Customers with social attributes (willing to share to others).
3. The card of the line is not bound before, and the customer of the card of the line is bound after the ticket is issued (the binding amount is increased).
The activation form in the embodiment of the invention: traditional financial institutions reach the best Hui Quan in the forms of short message pushing, log-in screen flicking, APP message pushing, APP intra-search and the like by issuing coupons on the APP of the respective institutions.
The method and the system aim to improve the binding rate of clients on public numbers of financial institutions, predict guest groups in a federal modeling mode, predict the predicted guest groups to be more willing to be stimulated by coupons to unbind cards and more willing to be shared to enable activities to form a chain effect, and then release coupons to the people in the touch mode.
The embodiment of the invention can be used for corresponding models: binding card prediction two-class model predicts that the client will bind after seeing the active message, y=0: after seeing the activity message, the client is not stimulated by the coupon and can not bind the bank card on the corresponding public number; y=1: after seeing the activity message, the customer binds the bank card to the corresponding public number.
1. First, a step of pre-accurate, i.e., a step of pre-federal learning modeling is introduced.
1. First, we introduce model Y value (binding probability) definition
According to the embodiment of the invention, the binding card preferential activity information can be touched in modes of short message pushing, login screen flicking, APP message pushing, APP internal search and the like, whether the binding card is carried out on the corresponding public number or not can be judged, and an Xgboost classification model can be adopted.
Y=0: after the customer sees the activity message, the bank card will not be bound to the corresponding public number.
Y=1: after the customer sees the activity message, the bank card is bound on the corresponding public number.
From the foregoing, in one embodiment, the binding probability prediction model may be an Xgboost model.
In specific implementation, the binding card probability prediction model can be an Xgboost model, so that the prediction accuracy of the model can be further improved.
2. Next, the sample data used for modeling will be described.
Data characteristics within the institution:
(1) customer base attribute data, as shown in Table one below:
list one
(2) Coupon attribute data, as shown in Table two below:
watch II
(3) Customer consumption attribute data, as shown in Table three below:
watch III
(4) Customer binding card attribute data, as shown in Table four below:
table four
| English name | Chinese name |
| IS_TIED_CARD_JD | Whether the card is bound on the channel A of the consumption platform before the card binding activity |
| IS_TIED_CARD_ZFB | Whether the card is bound on the C channel of the consumption platform before the card binding activity |
| IS_TIED_CARD_WX | Whether the card is bound on the B channel of the consumption platform before the card binding activity |
| IS_TIED_CARD_MT | Whether the card is bound on the F channel of the consumption platform before the card binding activity |
| IS_TIED_CARD_DD | Whether the card is bound on the D channel of the consumption platform before the card binding activity |
| IS_TIED_CARD_PDD | Whether the card is bound on the E channel of the consumption platform before the card binding activity |
(5) The customer browses the purchase fund financial attribute data as shown in table five below:
TABLE five
The modeling effect using only the organization data is shown in the following table six:
TABLE six
| English name | Meaning of | Value-book mechanism |
| accuracy | Accuracy rate of | 0.7349 |
| recall | Recall rate of recall | 0.7515 |
| auc_train | Auc on training set | 0.7725 |
| auc_test | Auc on test set | 0.7564 |
From the foregoing, in one embodiment, the feature data within the mechanism may include: the user basic attribute feature data, the coupon attribute feature data, the user consumption attribute feature data, the user binding card attribute feature data or any combination of the user browsing purchase fund financial attribute feature data.
Optimizing using federal modeling
The federal modeling is mainly to expand customer internet behavior labels provided by external internet companies, and by combining existing data features of the organization, the customer consumption demands of financial institutions are deeply insight through various modeling modes. Providing a grip for matching a more accurate and efficient marketing solution.
Expanding federal modeling data features
Based on the existing data features of the institution (see description of feature data in the institution above for details), the extended financial feature dimension (internet behavior feature data) is about 2w+, and relates to securities, loans, insurance, banks, funds, futures, and the like. The characteristics include:
User app behavioral profile data: app usage statistics within 180 days of a user mainly comprise app diversity statistics features, frequency statistics features and behavior change statistics features;
interest tag feature data: the interest labels are mined according to the user access and installation behaviors;
ID mapping feature data: full network ID information associated therewith, including imei, idfa, cookie, etc.;
device information feature data: including os, model, brand, price, and device ratings;
deep network characteristic data: a DNN-based depth network and outputting an intermediate layer as a feature;
the office type characteristic data is divided into: office software, cloud disk, note memo, mailbox, business recruitment integration, part-time employment practice, internet recruitment, enterprise information inquiry, enterprise internal office, CRM marketing tools, personnel financial tools, enterprise application others, sales promotion management, sales exhibition industry assistant, work assistant, business office others;
the efficiency class feature data is divided into: wallpaper theme desktop, garbage cleaning, input method fonts, browser, networking WiFi, battery power saving, security encryption, root system, file management, application store, search, system tools and other, VPN, and splitting;
The reading type characteristic data are divided into: the novel cartoon is comprehensive, nationwide, daily diffuse, book listening FM, reader and other reading functions;
the shooting beautifying feature data are divided into: photographing, camera photographing, image beautifying, album gallery, video and audio editing and expression head portrait;
the video entertainment characteristic data is divided into: short video, live broadcast synthesis, game live broadcast, outdoor live broadcast of sports, movie episode; video synthesis, two-dimensional animation, european drama, japanese and Korean drama, variety, video player download, sports program, network television, box, online music synthesis, sleep music, DJ, music player download, K song, FM radio synthesis, music FM, listening book FM, vehicle-mounted FM, ringtone, musical instrument, phase-sound drama comment, video-audio entertainment others, amplification and song identification;
the social communication characteristic data are divided into: chat communication, community friend-making synthesis, co-city local communities, sports communities, picture communities, knowledge learning communities, dating friends, lovers, game communities, job-site communities, technical communication communities, star-chasing rice ball, interest tide communities, wedding affinity, expression head portraits, telephone communication, constellation fortune, vermicelli community management and social communication others;
The shopping price characteristic data are divided into: comprehensive electronic commerce, fresh electronic commerce, mother and infant electronic commerce, sea wash luxury, sports tide electronic commerce, beauty electronic commerce, literature and play technology electronic commerce, clothing shoe electronic commerce, wine electronic commerce, other electronic commerce, movie ticket play ticket, retail O2O local, second hand transaction, shopping guide, group purchase, return and saving money, member commission direct sale, television shopping and class B electronic commerce micro-commerce;
the news information feature data is divided into: comprehensive information, sports information, entertainment information, financial information, military information, automotive information, technical information, game information, local synthesis, deep foreign language, and fun segments;
the network earning characteristic data are divided into: block chain earning, part-time network earning, news network earning, sports network earning, game network earning, video network earning;
the convenient life characteristic data are divided into: catering take-out, food menu, mood diary hand account, weather, calendar, intelligent life, intelligent household furniture, security monitoring, community service, local living service integration, cleaning, washing, maintenance, receiving and sending express, business hall, renting house and buying house integration, renting house, buying house, finishing service, goods pulling and moving, household administration, government service integration, medical insurance social insurance, pet life, wedding service, life assistant integration, driver integration, net appointment, freight, driving, crowd-sourced running leg riding, express delivery collecting post station, freight logistics and house; intermediaries, property administrators, recruits;
The automobile characteristic data are divided into: service behind a car, violation inquiry, car information, driving examination, car lending insurance, car business, second-hand car, motorcycle, truck, other high-speed parking, refueling and charging, car owner and intelligent application;
the travel navigation characteristic data are divided into: map navigation, positioning, network taxi, sharing single vehicle, traffic ticket, bus subway and other high-speed parking;
the characteristic data of the tourism accommodation category is divided into: short renters, hotels, air travel Cheng Chushou, out-of-home, tour guide attractions, tourist mall communities, tourist accommodations, and others;
the financial investment characteristic data are divided into: the financial management platform, financial information, stock stir-frying software, port stranding software, stock allocation, stock recommendation software, simulated stock stir-frying, fund, coin stir-frying, multi-product synthesis, precious metals, futures, foreign exchange, trust, crude oil and communication learning forums;
the financial lending feature data is divided into: internet loan, internet finance, house loan, other business enterprise services, credit cards, payments, consumer finance, town commercial banks, village banks, stock-making commercial banks, national commercial establishments, rural commercial banks.
From the foregoing, in one embodiment, the feature data of the external internet company may be internet behavior feature data of the external internet company.
In particular implementations, the feature data within the mechanism may include: one or any combination of user basic attribute feature data, coupon attribute feature data, user consumption attribute feature data, user binding card attribute feature data or user browsing purchase fund financial attribute feature data; the characteristic data of the external internet company can be internet behavior characteristic data of the external internet company, and the prediction accuracy of the model can be further improved.
3. Step of federal modeling:
in one embodiment, as shown in fig. 2, the coupon pushing method based on federal modeling may further include: the binding card probability prediction model is built in advance according to the following method of federal learning:
step 201: acquiring characteristic data of a historical user in a financial institution serving as a data initiator, binding probability corresponding to each characteristic data, and user identification encrypted in a preset encryption mode;
step 202: acquiring characteristic data of an external internet company of a historical user as a data service party and a user identifier encrypted in the preset encryption mode;
step 203: on the federal modeling platform, performing database collision and intersection on user identification encrypted in a preset encryption mode to obtain characteristic data of historical users existing in both a data initiator and a data service party;
Step 204: and carrying out federal learning modeling by utilizing the binding probability corresponding to the characteristic data of the data initiator by utilizing the characteristic data of the historical user existing in both the data initiator and the data server to obtain the binding probability prediction model.
In specific implementation, the implementation mode of establishing the binding card probability prediction model in advance by the federal learning can further improve the prediction accuracy of the model. The following is a detailed description.
1. A step of acquiring sample data for federal modeling, namely, the above steps 201 to 202: the two parties prepare data, a financial institution (such as a bank) side is defined as a data initiator A, an external internet company is defined as a data service side B, the data initiator prepares Y values (binding probability), and all characteristic data and encrypted mobile phone numbers mentioned by 'sample data used for modeling' above data characteristics prepared by the data initiator; the data service side does not need to prepare the Y value, but only needs to prepare all the characteristic data and the encrypted mobile phone number mentioned in the sample data used for modeling; in order to reduce the amount of data to be prepared and to increase the pool hit rate of both parties, both parties may define the area in advance, for example, the area a, and both parties prepare only the data of the users in the area a.
As can be seen from the foregoing, in one embodiment, on the federal modeling platform, performing a database collision and intersection with a user identifier encrypted in a preset encryption manner to obtain feature data of a historical user existing in both a data initiator and a data server may include: and on the federal modeling platform, the user identification encrypted in a preset encryption mode is subjected to library collision and intersection in a preset area (for example, area A) to obtain the characteristic data of the historical users existing in both the data initiator and the data service party.
2. A step of bumping into a warehouse, namely the step 203: the two parties perform the database collision and intersection solving by using the mobile phone number encrypted in the same way on the federal modeling platform (the federal belongs to the longitudinal federation, the longitudinal federation is equivalent to the characteristic of an extended user, the database collision and intersection solving is the operation of unifying user IDs and the operation of intersection solving of two data sets), and the data of the users existing in the two parties are obtained. In one embodiment, the user identifier encrypted in the preset encryption mode may be a user mobile phone number encrypted in the preset encryption mode, so that the efficiency of the database collision and the intersection can be improved, and the modeling efficiency is further improved.
3. Modeling step, namely step 104 described above:
Modeling related data and parameters: data initiator a: θ_a: a-party parameters (data initiator model parameters), x_a: a-party data (characteristic data of the data initiator), Y: a party Y value (binding probability).
Data service side B: θ_b: b-party parameters (data server model parameters), x_b: b-party data (characteristic data of the data service party).
The parameters are parameters of the model, both parties. The data is also data which are available to both parties, and the data of the other party cannot be seen by both parties, so that the data privacy can be protected.
In one embodiment, performing federal learning modeling to obtain the binding probability prediction model by using the binding probability corresponding to the feature data of the data initiator by using the feature data of the historical user existing in both the data initiator and the data server may include:
a data initiator (financial institution, such as a bank) generates a pair of public and private keys for homomorphic encryption, and sends the public keys to a data service party (external internet company);
and taking the initialized initiator model parameters and the initialized server model parameters as initial values, and circularly executing the following steps of updating the model parameters for a plurality of times until the optimal initiator model parameters and the optimal server model parameters are obtained:
The initiator determines the initiator model parameter of the current round, and obtains the product of the initiator model parameter (theta_A) of the current round and the characteristic data (X_A) of the initiator; the server determines the server model parameters of the current round, obtains the product of the server model parameters (theta_B) of the current round and the characteristic data (X_B) of the server, and sends the product to the initiator; the parameters circularly used in the first round are initialized initiator model parameters and initialized server model parameters;
the initiator obtains federal learning fusion model parameters (theta_X) of the current round according to the product; obtaining nonlinear function parameters (hθ (x)) of the current round according to the fused model parameters, and obtaining gradients (y-hθ (x)) of the model parameters of the initiator of the current round according to binding probabilities (y) corresponding to the characteristic data of the initiator and the nonlinear function parameters; encrypting the gradient of the initiator model parameter of the current round by utilizing the public key to obtain the encrypted gradient ([ y-hθ (x) ]) of the initiator model parameter of the current round, and transmitting the gradient to a server;
the server obtains the gradient ([ theta_B gradient ]) of the encrypted current round of server model parameters according to the gradient of the encrypted current round of initiator model parameters and the characteristic data of the server; generating a random number (R), and encrypting the random number by using a public key to obtain an encrypted random number ([ R ]); the gradient of the server model parameter of the current round after encryption and the random number after encryption are sent to an initiator;
The initiator decrypts the encrypted gradient of the current round service side model parameter and the encrypted random number by using the private key to obtain the decrypted gradient of the current round service side model parameter and the decrypted random number, and sends the decrypted current round service side model parameter and the decrypted random number to the data service side;
the data server removes the random number to obtain the gradient of the model parameters of the current round server;
the data initiator obtains updated initiator model parameters according to the gradients of the initiator model parameters of the current round; the data server obtains updated server model parameters according to the gradient of the current round server model parameters;
when a preset cycle termination condition is met, obtaining optimal initiator model parameters and server model parameters, wherein the optimal initiator model parameters and server model parameters are used for establishing the binding card probability prediction model; and repeatedly executing the step of updating the model parameters when the preset cycle termination condition is not met.
In the specific implementation, the specific implementation mode of the federal learning pre-established binding card probability prediction model can further improve the prediction accuracy of the model on the premise of protecting the safety and privacy of data. The modeling procedure is shown in table seven below.
Watch seven
In Table seven above, h (x) is a nonlinear function in the model, such as a sigmoid function, increasing the nonlinearity of the model, and hθ (x) is a nonlinear function parameter. θ_x is a model parameter, because privacy needs to be protected, after federal learning is adopted, θ_x needs to be determined after A, B two sides are combined, and the model parameter is the federal learning fusion model parameter.
2. Next, the steps actually applied after completion of the federal learning modeling, that is, the above steps 101 to 103 are described.
In the above step 101, feature data of the user to be predicted in the financial institution and feature data of the external internet company are acquired. In the step 102, the obtained feature data is input into the binding probability prediction model established by the federal learning in advance, so as to obtain the binding probability of each user to be predicted on the public number of the financial institution. A step of accurate delivery, namely the step 103: the federal modeling is utilized to combine the built model, the accuracy is higher, the output value is the probability of the client binding card, the maximum value is 1, the minimum value is 0, the client binding card probability is ordered from big to small, the client with higher predicted binding card probability can be subjected to coupon release (for example, the client with higher binding card probability than 0.95 is subjected to coupon release), the channel of coupon release can be search in a mobile phone banking app, sending a coupon short message, a mobile phone banking app popup window and the like, and after the client goes to the binding card on the public number of the consumption platform B, the consumption platform B immediate gold reduction coupon can be obtained.
As can be appreciated from the foregoing, in one embodiment, as shown in FIG. 3, pushing coupons to the most suitable group of users for issuance based on the probability of binding the user to be predicted to the financial institution public number may comprise:
step 1031: sorting the card binding probability of the user to be predicted on the public number of the financial institution from large to small;
step 1032: pushing the coupons to a user group with a binding probability higher than a preset value.
In the implementation, the specific implementation mode of pushing the coupons to the user groups most suitable for being issued can further improve the accuracy and efficiency of pushing the coupons.
Model one effect improvement after federal modeling is shown in the following table eight:
table eight
| English name | Meaning of | Value-book mechanism | Value-federation | Lifting up |
| accuracy | Accuracy rate of | 0.7349 | 0.8219 | 8.70％ |
| recall | Recall rate of recall | 0.7515 | 0.8504 | 9.89％ |
| auc_train | Auc on training set | 0.7725 | 0.8214 | 4.89％25 |
| auc_test | Auc on test set | 0.7564 | 0.8132 | 5.68％ |
The embodiment of the invention also provides a coupon pushing device based on federal modeling, which is described in the following embodiment. Because the principle of the device for solving the problem is similar to that of the coupon pushing method based on federal modeling, the implementation of the device can be referred to the implementation of the coupon pushing method based on federal modeling, and the repetition is not repeated.
FIG. 4 is a schematic structural diagram of a coupon pushing device based on federal modeling according to an embodiment of the present invention, as shown in FIG. 4, the device includes:
an obtaining unit 01, configured to obtain feature data of a user to be predicted in a financial institution and feature data of an external internet company;
the prediction unit 02 is used for inputting the characteristic data of the user to be predicted in the financial institution and the characteristic data of the external internet company into a pre-established binding probability prediction model to obtain the binding probability of the user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company;
and the pushing unit 03 is used for pushing the coupons to the user groups most suitable for issuing according to the binding probability of the users to be predicted on the public numbers of the financial institutions.
In one embodiment, the coupon pushing method based on federal modeling may further include: the establishing unit is used for pre-establishing the binding card probability prediction model by federal learning according to the following method:
acquiring characteristic data of a historical user in a financial institution serving as a data initiator, binding probability corresponding to each characteristic data, and user identification encrypted in a preset encryption mode;
Acquiring characteristic data of an external internet company of a historical user as a data service party and a user identifier encrypted in the preset encryption mode;
on the federal modeling platform, performing database collision and intersection on user identification encrypted in a preset encryption mode to obtain characteristic data of historical users existing in both a data initiator and a data service party;
and carrying out federal learning modeling by utilizing the binding probability corresponding to the characteristic data of the data initiator by utilizing the characteristic data of the historical user existing in both the data initiator and the data server to obtain the binding probability prediction model.
In one embodiment, performing federal learning modeling to obtain the binding probability prediction model by using the binding probability corresponding to the feature data of the data initiator by using the feature data of the historical user existing in both the data initiator and the data server may include:
the data initiator generates a pair of public and private keys for homomorphic encryption and sends the public keys to the data server;
and taking the initialized initiator model parameters and the initialized server model parameters as initial values, and circularly executing the following steps of updating the model parameters for a plurality of times until the optimal initiator model parameters and the optimal server model parameters are obtained:
The initiator determines the initiator model parameter of the current round, and obtains the product of the initiator model parameter of the current round and the characteristic data of the initiator; the server determines the server model parameters of the current turn, obtains the product of the server model parameters of the current turn and the characteristic data of the server, and sends the product to the initiator;
the initiator obtains federal learning fusion model parameters of the current round according to the product; obtaining nonlinear function parameters of the current round according to the fused model parameters, and obtaining gradients of the model parameters of the initiator of the current round according to binding probability corresponding to the characteristic data of the initiator and the nonlinear function parameters; encrypting the gradient of the initiator model parameter of the current round by using the public key to obtain the encrypted gradient of the initiator model parameter of the current round, and transmitting the gradient to a server;
the server obtains the gradient of the encrypted current round of server model parameters according to the gradient of the encrypted current round of initiator model parameters and the characteristic data of the server; generating a random number, and encrypting the random number by using a public key to obtain an encrypted random number; the gradient of the server model parameter of the current round after encryption and the random number after encryption are sent to an initiator;
The initiator decrypts the encrypted gradient of the current round service side model parameter and the encrypted random number by using the private key to obtain the decrypted gradient of the current round service side model parameter and the decrypted random number, and sends the decrypted current round service side model parameter and the decrypted random number to the data service side;
the data server removes the random number to obtain the gradient of the model parameters of the current round server;
the data initiator obtains updated initiator model parameters according to the gradients of the initiator model parameters of the current round; the data server obtains updated server model parameters according to the gradient of the current round server model parameters;
when a preset cycle termination condition is met, obtaining optimal initiator model parameters and server model parameters, wherein the optimal initiator model parameters and server model parameters are used for establishing the binding card probability prediction model; and repeatedly executing the step of updating the model parameters when the preset cycle termination condition is not met.
In one embodiment, on the federal modeling platform, performing a database collision and intersection with a user identifier encrypted in a preset encryption manner to obtain feature data of a historical user existing in both a data initiator and a data server, where the feature data may include: and on the federal modeling platform, the user identification encrypted in a preset encryption mode is subjected to database collision and intersection in a preset area range to obtain the characteristic data of the historical users existing in both the data initiator and the data server.
In one embodiment, the user identifier encrypted in the preset encryption manner may be a user mobile phone number encrypted in the preset encryption manner.
In one embodiment, the feature data within the institution may include: one or any combination of user basic attribute feature data, coupon attribute feature data, user consumption attribute feature data, user binding card attribute feature data or user browsing purchase fund financial attribute feature data; the feature data of the external internet company may be internet behavior feature data of the external internet company.
In one embodiment, the pushing unit is specifically configured to:
sorting the card binding probability of the user to be predicted on the public number of the financial institution from large to small;
pushing the coupons to a user group with a binding probability higher than a preset value.
In one embodiment, the binding probability prediction model may be an Xgboost model.
Based on the foregoing inventive concept, as shown in fig. 5, the present invention further proposes a computer device 500, including a memory 510, a processor 520, and a computer program 530 stored on the memory 510 and executable on the processor 520, where the processor 520 implements the foregoing federal modeling-based coupon pushing method when executing the computer program 530.
The embodiment of the invention also provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program realizes the coupon pushing method based on federal modeling when being executed by a processor.
The embodiment of the invention also provides a computer program product, which comprises a computer program, wherein the computer program realizes the coupon pushing method based on federal modeling when being executed by a processor.
In the embodiment of the invention, compared with the technical scheme of low coupon pushing precision in the prior art, the coupon pushing scheme based on federal modeling is characterized by comprising the following steps: acquiring characteristic data of a user to be predicted in a financial institution and characteristic data of an external internet company; inputting the characteristic data into a binding probability prediction model to obtain the binding probability of a user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company; and pushing the coupons to the user groups which are most suitable for issuing according to the binding probability of the users to be predicted on the public numbers of the financial institutions. The invention can expand the characteristic data provided by external Internet companies, and predict the binding probability of users on the public numbers of financial institutions by combining the characteristic data of the institutions in a federal modeling mode, so as to push coupons to the user groups most suitable for issuing.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (12)
1. A federal modeling-based coupon pushing method, comprising:
acquiring characteristic data of a user to be predicted in a financial institution and characteristic data of an external internet company;
inputting the characteristic data of the user to be predicted in the financial institution and the characteristic data of the external internet company into a pre-established binding probability prediction model to obtain the binding probability of the user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company;
and pushing the coupons to the user groups which are most suitable for issuing according to the binding probability of the users to be predicted on the public numbers of the financial institutions.
2. The method as recited in claim 1, further comprising: the binding card probability prediction model is built in advance according to the following method of federal learning:
acquiring characteristic data of a historical user in a financial institution serving as a data initiator, binding probability corresponding to each characteristic data, and user identification encrypted in a preset encryption mode;
acquiring characteristic data of an external internet company of a historical user as a data service party and a user identifier encrypted in the preset encryption mode;
On the federal modeling platform, performing database collision and intersection on user identification encrypted in a preset encryption mode to obtain characteristic data of historical users existing in both a data initiator and a data service party;
and carrying out federal learning modeling by utilizing the binding probability corresponding to the characteristic data of the data initiator by utilizing the characteristic data of the historical user existing in both the data initiator and the data server to obtain the binding probability prediction model.
3. The method of claim 2, wherein performing federal learning modeling to obtain the binding probability prediction model using the binding probability corresponding to the feature data of the data initiator by using the feature data of the historical user existing in both the data initiator and the data server, comprises:
the data initiator generates a pair of public and private keys for homomorphic encryption and sends the public keys to the data server;
and taking the initialized initiator model parameters and the initialized server model parameters as initial values, and circularly executing the following steps of updating the model parameters for a plurality of times until the optimal initiator model parameters and the optimal server model parameters are obtained:
the initiator determines the initiator model parameter of the current round, and obtains the product of the initiator model parameter of the current round and the characteristic data of the initiator; the server determines the server model parameters of the current turn, obtains the product of the server model parameters of the current turn and the characteristic data of the server, and sends the product to the initiator;
The initiator obtains federal learning fusion model parameters of the current round according to the product; obtaining nonlinear function parameters of the current round according to the fused model parameters, and obtaining gradients of the model parameters of the initiator of the current round according to binding probability corresponding to the characteristic data of the initiator and the nonlinear function parameters; encrypting the gradient of the initiator model parameter of the current round by using the public key to obtain the encrypted gradient of the initiator model parameter of the current round, and transmitting the gradient to a server;
the server obtains the gradient of the encrypted current round of server model parameters according to the gradient of the encrypted current round of initiator model parameters and the characteristic data of the server; generating a random number, and encrypting the random number by using a public key to obtain an encrypted random number; the gradient of the server model parameter of the current round after encryption and the random number after encryption are sent to an initiator;
the initiator decrypts the encrypted gradient of the current round service side model parameter and the encrypted random number by using the private key to obtain the decrypted gradient of the current round service side model parameter and the decrypted random number, and sends the decrypted current round service side model parameter and the decrypted random number to the data service side;
The data server removes the random number to obtain the gradient of the model parameters of the current round server;
the data initiator obtains updated initiator model parameters according to the gradients of the initiator model parameters of the current round; the data server obtains updated server model parameters according to the gradient of the current round server model parameters;
when a preset cycle termination condition is met, obtaining optimal initiator model parameters and server model parameters, wherein the optimal initiator model parameters and server model parameters are used for establishing the binding card probability prediction model; and repeatedly executing the step of updating the model parameters when the preset cycle termination condition is not met.
4. The method of claim 2, wherein, on the federal modeling platform, the library-hit intersection is performed with the user identifier encrypted in a preset encryption manner to obtain feature data of the historical user existing in both the data initiator and the data server, comprising: and on the federal modeling platform, the user identification encrypted in a preset encryption mode is subjected to database collision and intersection in a preset area range to obtain the characteristic data of the historical users existing in both the data initiator and the data server.
5. The method of claim 2, wherein the user identification encrypted in the predetermined encryption manner is a user handset number encrypted in the predetermined encryption manner.
6. The method of claim 1, wherein the feature data within the institution comprises: one or any combination of user basic attribute feature data, coupon attribute feature data, user consumption attribute feature data, user binding card attribute feature data or user browsing purchase fund financial attribute feature data; the characteristic data of the external internet company is internet behavior characteristic data of the external internet company.
7. The method of claim 1, wherein pushing coupons to the most suitable group of users for distribution based on the probability of binding the user to be predicted to the financial institution's public number, comprising:
sorting the card binding probability of the user to be predicted on the public number of the financial institution from large to small;
pushing the coupons to a user group with a binding probability higher than a preset value.
8. The method of claim 1, wherein the binding probability prediction model is an Xgboost model.
9. A federally modeled coupon pushing device, comprising:
The acquiring unit is used for acquiring the characteristic data of the user to be predicted in the financial institution and the characteristic data of the external internet company;
the prediction unit is used for inputting the characteristic data of the user to be predicted in the financial institution and the characteristic data of the external internet company into a pre-established binding probability prediction model to obtain the binding probability of the user to be predicted on the public number of the financial institution; the binding probability prediction model is pre-established by federation learning according to the characteristic data of the historical user in the financial institution and the characteristic data of the external internet company;
and the pushing unit is used for pushing the coupons to the user groups which are most suitable for being issued according to the binding probability of the users to be predicted on the public numbers of the financial institutions.
10. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method of any of claims 1 to 8 when executing the computer program.
11. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program which, when executed by a processor, implements the method of any of claims 1 to 8.
12. A computer program product, characterized in that the computer program product comprises a computer program which, when executed by a processor, implements the method of any of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310386490.2A CN116523069A (en) | 2023-04-12 | 2023-04-12 | Coupon pushing method and device based on federal modeling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310386490.2A CN116523069A (en) | 2023-04-12 | 2023-04-12 | Coupon pushing method and device based on federal modeling |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116523069A true CN116523069A (en) | 2023-08-01 |
Family
ID=87398580
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310386490.2A Pending CN116523069A (en) | 2023-04-12 | 2023-04-12 | Coupon pushing method and device based on federal modeling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116523069A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116757286A (en) * | 2023-08-16 | 2023-09-15 | 杭州金智塔科技有限公司 | Multi-party joint causal tree model construction system and method based on federal learning |
-
2023
- 2023-04-12 CN CN202310386490.2A patent/CN116523069A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116757286A (en) * | 2023-08-16 | 2023-09-15 | 杭州金智塔科技有限公司 | Multi-party joint causal tree model construction system and method based on federal learning |
| CN116757286B (en) * | 2023-08-16 | 2024-01-19 | 杭州金智塔科技有限公司 | Multi-party joint causal tree model construction system and method based on federal learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Juska | Integrated marketing communication: advertising and promotion in a digital world | |
| Dobrev et al. | Dynamics of niche width and resource partitioning | |
| Ma | China's mobile economy: opportunities in the largest and fastest information consumption boom | |
| CN101981587A (en) | Analyzing transactional data | |
| WO2011072105A1 (en) | System and method of a purchasing information platform | |
| CN108628866A (en) | The method and apparatus of data fusion | |
| CN111951052B (en) | Method and device for acquiring potential clients based on knowledge graph | |
| Batra et al. | M-commerce in India: Emerging issues | |
| Curran | The internet of dreams: Reinterpreting the internet | |
| CN116523069A (en) | Coupon pushing method and device based on federal modeling | |
| US20180114236A1 (en) | Method for Predicting a Demand for Vehicles for Hire | |
| Bathla et al. | Applying Metalytics to Measure Customer Experience in the Metaverse | |
| Pearson | The loyalty leap: Turning customer information into customer intimacy | |
| Hauer | Society caught in a labyrinth of algorithms: disputes, promises, and limitations of the new order of things | |
| CN110503470B (en) | Operation method and system for product sales promotion | |
| CN117271787A (en) | Knowledge base construction, intelligent customer service realization and knowledge question-answering method, equipment and medium | |
| Ernawati et al. | Gojek's Strategy to Win the Online Transportation Competition | |
| Bansal | Growth of the electronic commerce in China and India: A comparative study | |
| Bansal | Prospects of electronic commerce in India | |
| CN111787042A (en) | Method and device for pushing information | |
| Aman et al. | Digital Marketing As A Driver For The Sustainable Tourism Development | |
| O’Farrell | Developments in online, social media marketing in China and the West: an overview of different approaches | |
| CN107656963A (en) | Vehicle owner identification method and device, computer equipment and computer-readable recording medium | |
| Zheng et al. | Fundamentals of E-commerce | |
| Arief et al. | Comparative Analysis of Gojek and Grab Excellence Strategies in Improving the Market in Jember Regency |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |