CN113763019A - User information management method and device - Google Patents
User information management method and device Download PDFInfo
- Publication number
- CN113763019A CN113763019A CN202110117719.3A CN202110117719A CN113763019A CN 113763019 A CN113763019 A CN 113763019A CN 202110117719 A CN202110117719 A CN 202110117719A CN 113763019 A CN113763019 A CN 113763019A
- Authority
- CN
- China
- Prior art keywords
- user
- item
- promotion
- model
- evaluation model
- 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
- 238000007726 management method Methods 0.000 title claims abstract description 32
- 238000013210 evaluation model Methods 0.000 claims abstract description 129
- 238000012549 training Methods 0.000 claims abstract description 74
- 238000003062 neural network model Methods 0.000 claims abstract description 55
- 238000000034 method Methods 0.000 claims abstract description 35
- 238000006243 chemical reaction Methods 0.000 claims description 31
- 230000001737 promoting effect Effects 0.000 claims description 28
- 238000000605 extraction Methods 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 8
- 239000000126 substance Substances 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 7
- 230000006870 function Effects 0.000 description 15
- 238000010586 diagram Methods 0.000 description 10
- 230000004044 response Effects 0.000 description 8
- 230000006399 behavior Effects 0.000 description 6
- 238000004891 communication Methods 0.000 description 6
- 230000001364 causal effect Effects 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 230000035945 sensitivity Effects 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 239000000835 fiber Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- 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
- G06Q30/0224—Discounts or incentives, e.g. coupons or rebates based on user history
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06Q30/0283—Price estimation or determination
-
- 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
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0641—Shopping interfaces
Abstract
The invention discloses a user information management method and device, and relates to the technical field of computers. One embodiment of the method comprises: the method comprises the steps that a neural network model is trained by utilizing promotion scene user data according to promotion features of an article combined by a plurality of promotion means, the article features and corresponding user features, and a first evaluation model is generated; training a neural network model by using non-promotion scene user data to generate a second evaluation model; determining a gain prediction model based on the difference value of the output values of the first evaluation model and the second evaluation model, and predicting a gain value obtained by a user based on the item promotion characteristics through the gain prediction model; the accuracy of managing the user information according to the article sales promotion features is improved, and the accuracy of evaluating the article sales promotion effect is further improved.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a method and an apparatus for managing user information.
Background
At present, artificial intelligence technology is widely applied in the field of electronic commerce, and intelligent marketing is realized through the artificial intelligence technology; the existing intelligent marketing model is mainly based on the prediction of the click rate of a user, the total volume of commodity trades and the like; since the primary goal of promotions is to achieve maximum benefit with a set cost investment, intelligent marketing models are also typically used to manage user information to more accurately find target users that are sensitive to promotions.
However, in the existing mode of managing user information by using an intelligent marketing model constructed based on the click rate of the user and the total amount of commodity trades, the user can only be managed by predicting the click probability or purchase probability of the user. It is difficult to determine whether a user has made a purchase because of a promotion method, and thus there is a problem in that the accuracy of management and prediction of results is low for a group of conversion users corresponding to a promotion.
Disclosure of Invention
In view of this, embodiments of the present invention provide a method and an apparatus for managing user information, which can train a neural network model by using promotion scenario user data according to a promotion feature of an article combined by multiple promotion means, in combination with the article feature and a corresponding user feature, to generate a first evaluation model; training a neural network model by using non-promotion scene user data to generate a second evaluation model; determining a gain prediction model based on the difference value of the output values of the first evaluation model and the second evaluation model, and predicting a gain value obtained by a user based on the item promotion characteristics through the gain prediction model; the accuracy of managing the user information according to the article sales promotion features is improved, and the accuracy of evaluating the article sales promotion effect is further improved.
In order to achieve the above object, according to an aspect of an embodiment of the present invention, there is provided a user information management method, including: acquiring user historical data, and dividing the user historical data into promotion scene user data and non-promotion scene user data according to the operation condition of a user on an article display page and promotion characteristics of an article displayed on the article display page, wherein the operation condition of the user on the article display page is included in the user historical data; training a neural network model by using the promotion scene user data to generate a first evaluation model; training a neural network model by using the non-promotion scene user data to generate a second evaluation model; inputting user data to be predicted into the first evaluation model and the second evaluation model respectively to obtain an output value of the first evaluation model and an output value of the second evaluation model; calculating a gain value of the user to be predicted based on a difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion feature and the item feature of the item display page operated by the user to be predicted, which are included in the data of the user to be predicted; and managing the user to be predicted according to the gain value.
Optionally, the user information management method, characterized in that,
training a neural network model using the promotion scenario user data, comprising: training a neural network model by using user characteristics of a plurality of users, article characteristics of an article display page operated by the users, article promotion characteristics and operation conditions of the users on the article display page, wherein the user characteristics comprise the promotion scene user data; and/or training a neural network model using the non-promotional scenario user data, comprising: and training a neural network model by using the user characteristics of a plurality of users, the item characteristics of the item display page operated by the users and the operation condition of the item display page operated by the users, wherein the user characteristics of the users, the item characteristics of the item display page operated by the users and the operation condition of the users on the item display page are included in the non-promotion scene user data.
Optionally, the user information management method is characterized by further comprising: constructing a gain prediction model by using the difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion characteristics and the item characteristics of the item display page operated by the user to be predicted, which are included in the user data to be predicted; and executing the step of calculating the gain value of the user to be predicted by utilizing the gain prediction model.
Optionally, the user information management method is characterized by further comprising: obtaining a positive sample and a negative sample, wherein the positive sample comprises user data that is sensitive to an upsell feature of the item and the negative sample comprises user data that is not sensitive to an upsell feature of the item; adjusting the gain prediction model using the positive samples and the negative samples; and executing the step of calculating the gain value of the user to be predicted by utilizing the adjusted gain prediction model.
Optionally, the method for managing user information, wherein adjusting the gain prediction model includes: generating a training label by combining the operation condition of the user on the article display page and the condition that the user is sensitive to the promotion feature of the article, wherein the operation condition of the user on the article display page is included in the positive sample and the negative sample; and adjusting the gain prediction model by utilizing the positive sample, the negative sample, the training label corresponding to the positive sample and the training label corresponding to the negative sample.
Optionally, the user information management method, characterized in that,
the article promotion features are generated by combining a plurality of article promotion modes.
Optionally, the user information management method is characterized by further comprising:
when the neural network model is trained, the item price of the item displayed on the item display page is trained, and the target price of the item is determined according to the training result of the item price.
Optionally, the user information management method, characterized in that,
the neural network model includes: an input layer, a plurality of hidden layers, an output layer;
the training neural network model comprises: the input layer includes the promotional contextual user data or the non-promotional contextual user data; the plurality of hidden layers perform feature extraction and feature classification based on the promotion scene user data or the non-promotion scene user data, and obtain an output layer based on the result of the feature extraction and the result of the feature classification; for the first assessment model, the output layer outputting a first purchase conversion rate corresponding to a promotional feature of the item, the first purchase conversion rate being an output value of the first assessment model; for the second evaluation model, the output layer outputs a second purchase conversion rate, which is an output value of the second evaluation model.
In order to achieve the above object, according to a second aspect of an embodiment of the present invention, there is provided a user information management apparatus, comprising: the device comprises a historical data acquisition module, a first evaluation model generation module, a second evaluation model generation module and a user gain value calculation module; wherein the content of the first and second substances,
the historical data acquisition module is used for acquiring historical data of a user, and dividing the historical data of the user into promotion scene user data and non-promotion scene user data according to the operation condition of the user on an article display page and promotion characteristics of an article displayed on the article display page, wherein the operation condition of the user on the article display page is included in the historical data of the user;
the generation first evaluation model module is used for training a neural network model by using the promotion scene user data to generate a first evaluation model;
the generation second evaluation model module is used for training a neural network model by using the non-promotion scene user data to generate a second evaluation model;
the user gain value calculation module is used for inputting user data to be predicted into the first evaluation model and the second evaluation model respectively to obtain an output value of the first evaluation model and an output value of the second evaluation model; calculating a gain value of the user to be predicted based on a difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion feature and the item feature of the item display page operated by the user to be predicted, which are included in the data of the user to be predicted; and managing the user to be predicted according to the gain value.
Optionally, the user information management apparatus, characterized in that,
the user gain value calculation module is further used for constructing a gain prediction model by using the difference value between the output value of the first evaluation model and the output value of the second evaluation model, the item promotion characteristics of the item display page operated by the user to be predicted and the item characteristics, which are included in the user data to be predicted; and executing the step of calculating the gain value of the user to be predicted by utilizing the gain prediction model.
Optionally, the user information management apparatus, characterized in that,
the module for calculating a user gain value is further configured to obtain a positive sample and a negative sample, wherein the positive sample comprises user data that is sensitive to an upsell feature of the item, and the negative sample comprises user data that is not sensitive to an upsell feature of the item; adjusting the gain prediction model using the positive samples and the negative samples; and executing the step of calculating the gain value of the user to be predicted by utilizing the adjusted gain prediction model.
To achieve the above object, according to a third aspect of the embodiments of the present invention, there is provided an electronic device for user information management, including: one or more processors; storage means for storing one or more programs which, when executed by the one or more processors, cause the one or more processors to carry out a method as claimed in any one of the above-mentioned user information management methods.
To achieve the above object, according to a fourth aspect of embodiments of the present invention, there is provided a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method as in any one of the above user information management methods.
One embodiment of the above invention has the following advantages or benefits: predicting a gain value obtained by a user based on the item promotion feature through a gain prediction model; the accuracy of managing the user information according to the article sales promotion features is improved, and the accuracy of evaluating the article sales promotion effect is further improved.
Further effects of the above-mentioned non-conventional alternatives will be described below in connection with the embodiments.
Drawings
The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein:
fig. 1 is a flowchart illustrating a user information management method according to a first embodiment of the present invention;
fig. 2 is a flowchart illustrating a user information management method according to a second embodiment of the present invention;
FIG. 3 is a schematic diagram of a gain prediction model generation provided by an embodiment of the invention;
fig. 4 is a schematic structural diagram of a user information management apparatus according to an embodiment of the present invention;
FIG. 5 is an exemplary system architecture diagram in which embodiments of the present invention may be employed;
fig. 6 is a schematic block diagram of a computer system suitable for use in implementing a terminal device or server of an embodiment of the invention.
Detailed Description
Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, in which various details of embodiments of the invention are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
Currently, a gain Model (Uplift Model) can be used to predict the causal effect of a certain intervention on an individual's behavior, and thus the increment due to the intervention. Let us assume that X represents a user characteristic, Y ═ 1 represents a user's forward behavior (e.g., click or purchase), and G represents an intervention means (e.g., a coupon promotion included in a promotion characteristic), then the response model is expressed as: p (Y ═ 1| X), the gain model is expressed as: p (Y ═ 1| G, X). Therefore, the gain model can predict the promotion sensitivity of the user, thereby avoiding the waste of marketing cost. Common gain model modeling methods include:
differential response model: the differential response model is used for independently modeling the data of the experimental group and the control group in the AB experiment, and corresponding gain scores can be obtained by subtracting the scores of the two models in prediction. The difference model is simple to implement and is a common basic method of the gain model, and is also called T-learning.
X-learning: the X-learning algorithm is proposed on the basis of a T-learning model, and is based on the idea of using observed sample results and predicted unobserved sample results to carry out approximate learning on increments. Meanwhile, the X-learning can also adjust the weight of the result in a tendency way so as to achieve the purpose of optimizing the approximate result.
Direct modeling: direct modeling refers to direct modeling of gains by modifying an existing learner structure, and a common way is to modify a feature splitting method in a tree model. Gain is used as the main information gain part in node splitting, so that promotion sensitive people can be better distinguished.
However, the gain model applied to the e-commerce field mainly has the following problems:
the differential response model is formed by independently constructing two models, errors generated by the two models can be accumulated on a final result, and the self-modeling target is still a click conversion target in the response model and lacks the identification capability of a gain part.
The improved differential response model is trained in one model at a data level, but cannot meet the requirement that the user characteristics are independent of the conditional strategy.
The method for modifying the tree model is a method directly aiming at gain modeling, but because the loss function and the pruning need to be modified, the model convergence and the generalization are poor.
In view of this, an embodiment of the present invention provides a user information management method as shown in fig. 1, where the method may include the following steps:
step S101: acquiring user historical data, and dividing the user historical data into promotion scene user data and non-promotion scene user data according to the operation condition of a user on an article display page and promotion characteristics of an article displayed on the article display page, wherein the operation condition of the user on the article display page is included in the user historical data.
Specifically, user history data in a set time range is acquired, wherein the user history data comprises user characteristics, and the user characteristics comprise user portrait data (such as user age, gender, user occupation, purchase quantity, quantity of browsed articles, expense data) and the like; further, the operation condition of the user on the article display page and the promotion feature of the article displayed on the article display page are indicated, wherein the operation condition is the operation of the user on the article display page, such as: browsing items contained in the item display page, purchasing items contained in the item display page, etc., wherein the user browses or purchases items containing item features such as: the category, price, color, model, etc. of the item; the promotion feature of the item is a promotion mode for the item; for example: whether to use the coupon, whether to discount, whether to give a gift, whether to reduce, etc. obtain the promotion feature vector after encoding, for example: 1,0,1 … where 1 indicates that the item has the promotional method. That is, the item promotion feature is generated by combining a plurality of item promotion methods. The combined article promotion feature overcomes the defect that only promotion is judged in the existing improved differential response model, and improves the accuracy of predicting the promotion effect.
Further, according to the sales promotion features of the articles displayed on the article display page corresponding to the user, dividing the user historical data into sales promotion scene user data and non-sales promotion scene user data; if the articles (the articles displayed on the article display page) browsed or purchased by the user have the promotion features (for example, the promotion mode is any one or more of coupon use, discount use, gift giving and full reduction), the corresponding user history data is promotion scene user data; and otherwise, the historical data is non-promotion scene user data, namely, the historical user data is divided into promotion scene user data and non-promotion scene user data.
Step S102: and training a neural network model by using the promotion scene user data to generate a first evaluation model.
Specifically, training a neural network model using the promotion scenario user data includes: training a neural network model by using user characteristics (such as user age, gender, user occupation, purchase quantity, quantity of browsed items, expense data and the like) of a plurality of users included in the promotion scene user data, item characteristics (such as item types, prices, colors, models and the like) of an item display page operated by the users, item promotion characteristics (such as whether to use a coupon, whether to discount, whether to give a gift, whether to be full or not and the like) and operation conditions (such as browsing or purchasing and the like) of the users on the item display page;
further, the training neural network model firstly takes the user characteristics, the item characteristics and the item promotion characteristics as input samples of the training neural network model, wherein the input samples further comprise operation conditions (such as purchase or non-purchase and the like) of the user on an item display page, for example: the input samples are represented as: [ X ]1,X2,X3,…Y](ii) a Wherein, X1Representative user characteristics, X2Representing a characteristic of the article, X3Representing an item promotion feature (e.g., a feature vector as described in step S101); y represents a label of the user operation condition; training a Neural network model based on the respective input samples, wherein, preferably, DNN (Deep Neural Networks, DNN) is selected as the trained Neural network model; further, the first evaluation model generated based on the DNN comprises an input layer, a plurality of hidden layers, an output layer and a loss function corresponding to the output layer; namely, the neural network model includes: an input layer, a plurality of hidden layers, an output layer.
Further, the input layer includes the promotional scenario user data; the plurality of hidden layers perform feature extraction and feature classification based on the promotion scene user data or the non-promotion scene user data, and feature extraction and feature classification based on featuresObtaining an output layer according to the extracted result and the feature classification result; for the first assessment model, the output layer outputting a first purchase conversion rate corresponding to a promotional feature of the item, the first purchase conversion rate being an output value of the first assessment model; wherein the input layer contains training sample data containing the user characteristic, the item promotion characteristic, and a label indicating a user operation condition (e.g., a label indicating whether the user purchased the item); the input layer is represented as: a is0＝[X1,X2,X3,…Y]；
The plurality of hidden layers perform feature extraction and feature classification on the training sample data based on the training sample data; the hidden layer is represented as: a isn+1＝Relu(Wn*an+bn) Wherein n is a positive integer greater than or equal to 1, Relu is an activation function of the hidden layer, WnIs a matrix, bnFor offset vectors, it is understood that the hidden layer may be an N +1 layer, the input of the N +1 layer being the output of the nth layer, and so on.
Further, an output layer is determined from a tail layer of the hidden layer, i.e., based on the results of the feature extraction and the results of the feature classification, an output layer is derived that outputs a first purchase conversion rate of the item-promotion features of the item. For example: the tail layer of the hidden layer is the H-th layer, and can be represented as: p1 ═ WH*aH+bH(ii) a Training the DNN model to use a cross-entropy loss function, namely a loss function corresponding to the output layer, when determining the output layer, wherein P1 is a first purchase conversion rate corresponding to the promotion feature of the item, namely the first purchase conversion rate is an output value of the first evaluation model.
Step S103: and training a neural network model by using the non-promotion scene user data to generate a second evaluation model.
Specifically, training a neural network model using the non-promotion scenario user data includes: and training a neural network model by using the user characteristics of a plurality of users, the item characteristics of the item display page operated by the users and the operation condition of the item display page operated by the users, wherein the user characteristics of the users, the item characteristics of the item display page operated by the users and the operation condition of the users on the item display page are included in the non-promotion scene user data. The detailed description of the user characteristics, the article characteristics, and the operation conditions is consistent with the description of step S101 or step S102, and is not repeated here.
Further, the training neural network model firstly takes the user characteristics and the item characteristics as input samples of the training neural network model, the input samples further include operation conditions (for example: purchase or unpurctation and the like) of the user on the item display page, wherein the input samples of the training neural network model include the user characteristics, the item characteristics and the operation conditions (for example: a label indicating whether the user purchases the item) of the user on the item display page, such as: the input samples are represented as: [ X ]1,X2,…Y](ii) a Wherein, X1Representative user characteristics, X2A label representing a characteristic of an item, Y representing whether the user purchased the item;
training a Neural network model according to each input sample, preferably, selecting DNN (Deep Neural Networks, DNN) as the trained Neural network model; further, the second evaluation model generated based on the DNN includes an input layer, a plurality of hidden layers, and an output layer; the input layer includes the non-promotional scenario user data; the plurality of hidden layers perform feature extraction and feature classification based on the promotion scene user data or the non-promotion scene user data, and obtain an output layer based on the result of the feature extraction and the result of the feature classification; for the second evaluation model, the output layer outputs a second purchase conversion rate, which is an output value of the second evaluation model.
Wherein the input layer contains training sample data containing the user features, the item features, and a label indicating whether the user purchased the item; expressed as: c. C0＝[X1,X2,…Y]；
The plurality of hidden layers extract the features of the training sample data based on the training sample data andclassifying the characteristics; the hidden layer is represented as: c. Cn+1＝Relu(Wn*cn+bn) Wherein n is a positive integer greater than or equal to 1, Relu is an activation function of the hidden layer, WnIs a matrix, bnFor offset vectors, it is understood that the hidden layer may be an N +1 layer, the input of the N +1 layer being the output of the nth layer, and so on.
Further, an output layer is determined according to a tail layer of the hidden layer, namely, an output layer is obtained based on the result of the feature extraction and the result of the feature classification, the output layer outputs the item promotion feature of the item, and the output layer is used for generating a second purchase conversion rate corresponding to the item promotion feature of the user. For example: the tail layer of the hidden layer is the H-th layer, and can be represented as: p0 ═ WH*cH+bH(ii) a Training the DNN model to use a cross-entropy loss function when determining the output layer, namely, a loss function corresponding to the output layer, wherein P0 is a second purchase conversion rate corresponding to the promotional feature of the item, namely, the second purchase conversion rate is an output value of the second evaluation model.
As can be seen from the descriptions in step S102 and step S103, compared with the first evaluation model, the training sample of the first evaluation model is promotion scenario user data, and the training sample of the second evaluation model is non-promotion scenario user data, and the two evaluation models are used to compare the probability (i.e., conversion rate, i.e., user sensitivity to promotion) that the user purchases the item under the factors of the presence or absence of the promotion feature.
Further, a first task training neural network model is created, and a first evaluation model is generated; creating a second task training neural network model and generating a second evaluation model; the first task and the second task are synchronously performed; training a neural network model (such as DNN) in a multitask mode based on a condition average causal effect, and simultaneously generating a first evaluation model and a second evaluation model; when the model is generated, the obtained historical training data is incomplete, because the single user historical data sample in the training sample set cannot simultaneously obtain the user performances under the conditions of the influence of the promotion characteristic factors and the influence without the promotion characteristic factors, so that the causal effect expectation of all users (groups) is needed to be used for approximately evaluating the conversion rate of the whole user group under the promotion characteristic (namely the difference of the conversion rate under the promotion condition and the conversion rate under the non-promotion condition is the gain), and the effect is the conditional average causal effect.
Step S104: inputting user data to be predicted into the first evaluation model and the second evaluation model respectively to obtain an output value of the first evaluation model and an output value of the second evaluation model; calculating a gain value of the user to be predicted based on a difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion feature and the item feature of the item display page operated by the user to be predicted, which are included in the data of the user to be predicted; and managing the user to be predicted according to the gain value.
Specifically, according to the descriptions in step S102 and step S103, a first evaluation model and a second evaluation model are obtained, and user data to be predicted is respectively input into the first evaluation model and the second evaluation model, so as to obtain an output value of the first evaluation model (for example, P1, j: representing the jth user as the user to be predicted) and an output value of the second evaluation model (for example, P0, j: representing the jth user as the user to be predicted); further, the gain value of the user to be predicted is calculated based on the difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion feature and the item feature of the item display page operated by the user to be predicted, which are included in the data of the user to be predicted.
For example: calculating the gain value of the user to be predicted by taking the calculation of the input-output ratio of the articles corresponding to the plurality of users to be predicted as an example, as shown in formula (1), formula (2) is a constraint function of cost in formula (1):
wherein Δ ROI represents the sum of input-output ratios, P1,jAn output value representing a first assessment model for a jth user (e.g., a first purchase conversion rate); p0,jAn output value representing a second evaluation model for a jth user (e.g., a second purchase conversion rate); cost represents an item promotional characteristic of the item (e.g., the item promotional characteristic may be a promotional cost, such as multiplying a face value of a coupon by a probability of use of the coupon as a value of cost); budget represents the cost of the item as a constraint on the cost of the promotion (e.g., the cost of the promotion cannot be less than the cost of the item, or may be less than a set amount or percentage of the cost of the item, etc.).
Further, a gain prediction model is constructed by using the difference value of the output value of the first evaluation model and the output value of the second evaluation model (for example, based on the difference value of the first purchase conversion rate and the second purchase conversion rate), the item promotion feature of the item display page operated by the user to be predicted (for example, the cost contained in the formula (1)) and the item feature (for example, the item price, namely, the OrderPrice contained in the formula (1)) included in the user data to be predicted; and (2) utilizing the gain prediction model (taking the formula (1) as an example), and executing the step of calculating the gain value of the user to be predicted.
Further, a step of calculating a gain value of the user to be predicted is performed, specifically, the gain value of the user to be predicted is calculated and obtained through a gain prediction model (taking formula (1) as an example), so as to evaluate the sensitivity of the user to be predicted to the sales promotion features of the article, and further manage the user to be predicted, wherein the user information to be managed (i.e., the user information to be managed of the user to be predicted) may be user information classified, for example, into users sensitive to sales promotion, users insensitive to sales promotion, and the like, and may set corresponding labels and the like for the users according to the classification; wherein the user behavior of the promotion-sensitive user is to purchase if the item is promoted and not to purchase if the item is not promoted; the corresponding user behavior of a promotion-insensitive user is independent of whether the item has a promotional feature, i.e., whether purchasing or not purchasing the item is of low promotional relevance to the item. By managing the user information, the data basis for executing sales promotion aiming at the target population is obtained, and the accuracy of sales promotion of the articles is improved.
Further, different from the traditional response model, based on the gain prediction model, two groups of real mirror image users in the electronic mall abcd can be selected, the users in the experimental group are sorted according to the gain values estimated by the model, and finally the gain values of the two groups of users under different population quantiles are observed simultaneously. Under the same quantile, the difference value of the conversion rates of the experimental group and the control group is the gain value under the quantile, so that the accuracy of the gain prediction model is favorably verified.
As shown in fig. 2, an embodiment of the present invention provides a user information management method, which may include the following steps:
step S201: acquiring user historical data, and dividing the user historical data into promotion scene user data and non-promotion scene user data according to the operation condition of a user on an article display page and promotion characteristics of an article displayed on the article display page, wherein the operation condition of the user on the article display page is included in the user historical data; training a neural network model by using the promotion scene user data to generate a first evaluation model; and training a neural network model by using the non-promotion scene user data to generate a second evaluation model.
Specifically, the description of training the neural network model to generate the first evaluation model and the second evaluation model is consistent with steps S101 to S103, and is not repeated here.
Step S202: constructing a gain prediction model by using the difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion characteristics and the item characteristics of the item display page operated by the user to be predicted, which are included in the user data to be predicted; and executing the step of calculating the gain value of the user to be predicted by utilizing the gain prediction model.
Specifically, the specific description of the step of constructing the gain prediction model and calculating the gain value of the user to be predicted is consistent with step S104, and is not repeated here.
Step S203: obtaining a positive sample and a negative sample, wherein the positive sample comprises user data that is sensitive to an upsell feature of the item and the negative sample comprises user data that is not sensitive to an upsell feature of the item; adjusting the gain prediction model using the positive samples and the negative samples; and executing the step of calculating the gain value of the user to be predicted by utilizing the adjusted gain prediction model.
Specifically, the gain prediction model is further trained to improve the accuracy of the gain prediction model, and the method for training the gain prediction model comprises the following steps:
obtaining a positive sample and a negative sample, wherein the positive sample comprises user data that is sensitive to an upsell feature of the item and the negative sample comprises user data that is not sensitive to an upsell feature of the item; wherein the user behavior to which the promotional feature sensitivity corresponds is purchase if the item is promoted and not purchase if the item is not promoted; promotional characteristics insensitivity corresponding user behavior is independent of whether the item has a promotional characteristic, i.e., whether purchasing or not purchasing the item is of low promotional relevance to the item.
Further, using the positive samples and the negative samples, adjusting the gain prediction model; specifically, the gain prediction model is adjusted to optimize the gain prediction model, and the step of calculating the gain value of the user to be predicted is performed by using the adjusted gain prediction model. Further, generating a training label by combining the operation condition of the user on the article display page and the condition that the user is sensitive to the promotion feature of the article, wherein the operation condition of the user on the article display page is included in the positive sample and the negative sample; and adjusting the gain prediction model by utilizing the positive sample, the negative sample, the training label corresponding to the positive sample and the training label corresponding to the negative sample.
For example: the method for determining the training label comprises the following steps: as shown in the formula (3), yn is the operation condition of the user on the item display page (for example, whether the user purchases the item), ym is the condition that the user is sensitive to the promotion features of the item (for example, whether the user is sensitive to promotion), and further, the conversion rate yn corresponding to the first user data is combined; and generating a label delta ROI according to the conversion rate ym corresponding to the second user data, and adjusting the gain prediction model by using the positive sample, the negative sample, the training label corresponding to the positive sample and the training label corresponding to the negative sample (for example, the gain prediction model is shown as a formula (1)).
Further, the description of the step of calculating the gain value of the user to be predicted is consistent with the description of step S104 by using the adjusted gain prediction model, and is not repeated here.
As shown in fig. 3, an embodiment of the present invention provides a schematic diagram for constructing a gain prediction model, where:
the figure includes an input layer 301, a plurality of hidden layers 303, an output layer 303, and a gain prediction model 304.
Specifically, training a neural network model includes: the input layer 301 includes the promotional scene user data or the non-promotional scene user data; the hidden layers 302 perform feature extraction and feature classification based on the promotion scene user data or the non-promotion scene user data, and obtain the output layer 303 based on the result of the feature extraction and the result of the feature classification.
As shown in fig. 3, the promotion scenario user data includes user characteristics (user characteristic 1.. user characteristic n) of a plurality of users, item characteristics (item characteristic 1 … item characteristic n) of the item display page operated by the users, item promotion characteristics (item promotion characteristic 1 … item promotion characteristic n) and operation conditions of the item display page by the users; the non-promotional scenario user data includes user characteristics (user characteristics 1.. user characteristics n) of a plurality of users, item characteristics (item characteristics 1 … item characteristics n) of the item display page operated by the users, and the operation of the item display page by the users.
Further, training a neural network model, and obtaining an output layer by utilizing feature extraction and feature classification of a plurality of hidden layers; for the first assessment model, the output layer outputting a first purchase conversion rate corresponding to a promotional feature of the item, the first purchase conversion rate being an output value of the first assessment model; for the second evaluation model, the output layer outputs a second purchase conversion rate, which is an output value of the second evaluation model.
Constructing a gain prediction model by using the difference value of the output value of the first evaluation model and the output value of the second evaluation model, the item promotion feature of the item display page operated by the user to be predicted and the item feature included in the user data to be predicted (as shown in 304 of FIG. 3);
further, when the neural network model is trained, the item price of the displayed item on the item display page is trained, and the target price of the item is determined according to the training result of the item price. As shown in fig. 3, the output layer 303 includes an output value obtained by training the price of the item, that is, when the first evaluation model or the second evaluation model is generated, the item price of the item displayed on the item display page is determined, and the target price of the item is determined according to the training result of the item price. In particular, by training the price of an item, it is helpful for the provider of the item to intelligently price the item, including intelligently determining the manner and amount of promotions.
Further, based on the gain prediction model shown in 304, further optimization and adjustment are performed, and the description of the step of calculating the gain value of the user to be predicted is performed by using the adjusted gain prediction model and is consistent with the description of step S104, which is not repeated herein; that is, obtaining a positive sample including user data that is sensitive to an upsell feature of the item and a negative sample including user data that is not sensitive to an upsell feature of the item; adjusting the gain prediction model using the positive samples and the negative samples; and executing the step of calculating the gain value of the user to be predicted by utilizing the adjusted gain prediction model.
As shown in fig. 4, an embodiment of the present invention provides a user information management apparatus 400, including: a historical data acquisition module 401, a first evaluation model generation module 402, a second evaluation model generation module 403 and a user gain value calculation module 404; wherein the content of the first and second substances,
the historical data acquiring module 401 is configured to acquire user historical data, and divide the user historical data into promotion scene user data and non-promotion scene user data according to operation conditions of a user on an article display page and promotion features of an article displayed on the article display page, where the operation conditions include the user historical data;
the generate first evaluation model module 402 is configured to train a neural network model using the promotion scenario user data to generate a first evaluation model;
the generate second evaluation model module 403 is configured to train a neural network model using the non-promotion scene user data, and generate a second evaluation model;
the user gain value calculating module 404 is configured to input user data to be predicted into the first evaluation model and the second evaluation model respectively to obtain an output value of the first evaluation model and an output value of the second evaluation model; calculating a gain value of the user to be predicted based on a difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion feature and the item feature of the item display page operated by the user to be predicted, which are included in the data of the user to be predicted; and managing the user to be predicted according to the gain value.
Optionally, the module for calculating a user gain value 404 is further configured to construct a gain prediction model by using a difference between the output value of the first evaluation model and the output value of the second evaluation model, the item promotion feature of the item display page operated by the user to be predicted included in the user data to be predicted, and the item feature; and executing the step of calculating the gain value of the user to be predicted by utilizing the gain prediction model.
Optionally, the module for calculating a user gain value 404 is further configured to obtain a positive sample and a negative sample, wherein the positive sample includes user data sensitive to the promotional characteristic of the item, and the negative sample includes user data insensitive to the promotional characteristic of the item; adjusting the gain prediction model using the positive samples and the negative samples; and executing the step of calculating the gain value of the user to be predicted by utilizing the adjusted gain prediction model.
An embodiment of the present invention further provides an electronic device of a user information management apparatus, including: one or more processors; the storage device is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors are enabled to realize the method provided by any one of the above embodiments.
Embodiments of the present invention further provide a computer-readable medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method provided in any of the above embodiments.
Fig. 5 illustrates an exemplary system architecture 500 to which a user information management device method or user information management device of embodiments of the present invention may be applied.
As shown in fig. 5, the system architecture 500 may include terminal devices 501, 502, 503, a network 504, and a server 505. The network 504 serves to provide a medium for communication links between the terminal devices 501, 502, 503 and the server 505. Network 504 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.
The user may use the terminal devices 501, 502, 503 to interact with a server 505 over a network 504 to receive or send messages or the like. The terminal devices 501, 502, 503 may have various client applications installed thereon, such as an e-mall client application, a web browser application, a search-type application, an instant messaging tool, a mailbox client, and the like.
The terminal devices 501, 502, 503 may be various electronic devices having display screens and supporting various client applications, including but not limited to smartphones, tablets, laptop and desktop computers, and the like.
The server 505 may be a server providing various services, such as a background management server providing support for client applications used by users with the terminal devices 501, 502, 503. The background management server can process the received request for accessing the page object and feed back the promotion information corresponding to the object to the terminal equipment.
It should be noted that the user information management method provided in the embodiment of the present invention is generally executed by the server 505, and accordingly, the user information management apparatus is generally disposed in the server 505.
It should be understood that the number of terminal devices, networks, and servers in fig. 5 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
Referring now to FIG. 6, a block diagram of a computer system 600 suitable for use with a terminal device implementing an embodiment of the invention is shown. The terminal device shown in fig. 6 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.
As shown in fig. 6, the computer system 600 includes a Central Processing Unit (CPU)601 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM)602 or a program loaded from a storage section 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data necessary for the operation of the system 600 are also stored. The CPU 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.
The following components are connected to the I/O interface 605: an input portion 606 including a keyboard, a mouse, and the like; an output portion 607 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage section 608 including a hard disk and the like; and a communication section 609 including a network interface card such as a LAN card, a modem, or the like. The communication section 609 performs communication processing via a network such as the internet. The driver 610 is also connected to the I/O interface 605 as needed. A removable medium 611 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 610 as necessary, so that a computer program read out therefrom is mounted in the storage section 608 as necessary.
In particular, according to the embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 609, and/or installed from the removable medium 611. The computer program performs the above-described functions defined in the system of the present invention when executed by the Central Processing Unit (CPU) 601.
It should be noted that the computer readable medium shown in the present invention can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present invention, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present invention, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The modules and/or units described in the embodiments of the present invention may be implemented by software, and may also be implemented by hardware. The described modules and/or units may also be provided in a processor, and may be described as: a processor includes a module to obtain historical data, a module to generate a first evaluation model, a module to generate a second evaluation model, and a module to calculate a user gain value. The names of these modules do not limit the module itself in some cases, for example, the module for calculating the user gain value may be further described as a "module for calculating the user gain value based on the target gain prediction model according to the user characteristics and the article characteristics".
As another aspect, the present invention also provides a computer-readable medium that may be contained in the apparatus described in the above embodiments; or may be separate and not incorporated into the device. The computer readable medium carries one or more programs which, when executed by a device, cause the device to comprise: acquiring user historical data, and dividing the user historical data into promotion scene user data and non-promotion scene user data according to the operation condition of a user on an article display page and promotion characteristics of an article displayed on the article display page, wherein the operation condition of the user on the article display page is included in the user historical data; training a neural network model by using the promotion scene user data to generate a first evaluation model; training a neural network model by using the non-promotion scene user data to generate a second evaluation model; inputting user data to be predicted into the first evaluation model and the second evaluation model respectively to obtain an output value of the first evaluation model and an output value of the second evaluation model; calculating a gain value of the user to be predicted based on a difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion feature and the item feature of the item display page operated by the user to be predicted, which are included in the data of the user to be predicted; and managing the user to be predicted according to the gain value.
Predicting a gain value obtained by a user based on the item promotion feature through a gain prediction model; the accuracy of managing the user information according to the article sales promotion features is improved, and the accuracy of evaluating the article sales promotion effect is further improved.
The above-described embodiments should not be construed as limiting the scope of the invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions can occur, depending on design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (13)
1. A user information management method, comprising:
acquiring user historical data, and dividing the user historical data into promotion scene user data and non-promotion scene user data according to the operation condition of a user on an article display page and promotion characteristics of an article displayed on the article display page, wherein the operation condition of the user on the article display page is included in the user historical data;
training a neural network model by using the promotion scene user data to generate a first evaluation model;
training a neural network model by using the non-promotion scene user data to generate a second evaluation model;
inputting user data to be predicted into the first evaluation model and the second evaluation model respectively to obtain an output value of the first evaluation model and an output value of the second evaluation model;
calculating a gain value of the user to be predicted based on a difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion feature and the item feature of the item display page operated by the user to be predicted, which are included in the data of the user to be predicted; and managing the user to be predicted according to the gain value.
2. The method of claim 1,
training a neural network model using the promotion scenario user data, comprising:
training a neural network model by using user characteristics of a plurality of users, article characteristics of an article display page operated by the users, article promotion characteristics and operation conditions of the users on the article display page, wherein the user characteristics comprise the promotion scene user data;
and/or the presence of a gas in the gas,
training a neural network model using the non-promotional scenario user data, comprising:
and training a neural network model by using the user characteristics of a plurality of users, the item characteristics of the item display page operated by the users and the operation condition of the item display page operated by the users, wherein the user characteristics of the users, the item characteristics of the item display page operated by the users and the operation condition of the users on the item display page are included in the non-promotion scene user data.
3. The method of claim 1, further comprising:
constructing a gain prediction model by using the difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion characteristics and the item characteristics of the item display page operated by the user to be predicted, which are included in the user data to be predicted;
and executing the step of calculating the gain value of the user to be predicted by utilizing the gain prediction model.
4. The method of claim 3, further comprising:
obtaining a positive sample and a negative sample, wherein the positive sample comprises user data that is sensitive to an upsell feature of the item and the negative sample comprises user data that is not sensitive to an upsell feature of the item;
adjusting the gain prediction model using the positive samples and the negative samples;
and executing the step of calculating the gain value of the user to be predicted by utilizing the adjusted gain prediction model.
5. The method of claim 4, wherein adjusting the gain prediction model comprises:
generating a training label by combining the operation condition of the user on the article display page and the condition that the user is sensitive to the promotion feature of the article, wherein the operation condition of the user on the article display page is included in the positive sample and the negative sample;
and adjusting the gain prediction model by utilizing the positive sample, the negative sample, the training label corresponding to the positive sample and the training label corresponding to the negative sample.
6. The method of claim 1,
the article promotion features are generated by combining a plurality of article promotion modes.
7. The method of claim 1, further comprising:
when the neural network model is trained, the item price of the item displayed on the item display page is trained, and the target price of the item is determined according to the training result of the item price.
8. The method of claim 1,
the neural network model includes: an input layer, a plurality of hidden layers, an output layer;
the training neural network model comprises:
the input layer includes the promotional contextual user data or the non-promotional contextual user data;
the plurality of hidden layers perform feature extraction and feature classification based on the promotion scene user data or the non-promotion scene user data, and obtain an output layer based on the result of the feature extraction and the result of the feature classification;
for the first assessment model, the output layer outputting a first purchase conversion rate corresponding to a promotional feature of the item, the first purchase conversion rate being an output value of the first assessment model;
for the second evaluation model, the output layer outputs a second purchase conversion rate, which is an output value of the second evaluation model.
9. A user information management apparatus, comprising: the device comprises a historical data acquisition module, a first evaluation model generation module, a second evaluation model generation module and a user gain value calculation module; wherein the content of the first and second substances,
the historical data acquisition module is used for acquiring historical data of a user, and dividing the historical data of the user into promotion scene user data and non-promotion scene user data according to the operation condition of the user on an article display page and promotion characteristics of an article displayed on the article display page, wherein the operation condition of the user on the article display page is included in the historical data of the user;
the generation first evaluation model module is used for training a neural network model by using the promotion scene user data to generate a first evaluation model;
the generation second evaluation model module is used for training a neural network model by using the non-promotion scene user data to generate a second evaluation model;
the user gain value calculation module is used for inputting user data to be predicted into the first evaluation model and the second evaluation model respectively to obtain an output value of the first evaluation model and an output value of the second evaluation model; calculating a gain value of the user to be predicted based on a difference value between the output value of the first evaluation model and the output value of the second evaluation model, and the item promotion feature and the item feature of the item display page operated by the user to be predicted, which are included in the data of the user to be predicted; and managing the user to be predicted according to the gain value.
10. The prediction apparatus according to claim 9,
the user gain value calculation module is further used for constructing a gain prediction model by using the difference value between the output value of the first evaluation model and the output value of the second evaluation model, the item promotion characteristics of the item display page operated by the user to be predicted and the item characteristics, which are included in the user data to be predicted; and executing the step of calculating the gain value of the user to be predicted by utilizing the gain prediction model.
11. The prediction apparatus according to claim 10,
the module for calculating a user gain value is further configured to obtain a positive sample and a negative sample, wherein the positive sample comprises user data that is sensitive to an upsell feature of the item, and the negative sample comprises user data that is not sensitive to an upsell feature of the item; adjusting the gain prediction model using the positive samples and the negative samples;
and executing the step of calculating the gain value of the user to be predicted by utilizing the adjusted gain prediction model.
12. An electronic device, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-8.
13. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110117719.3A CN113763019A (en) | 2021-01-28 | 2021-01-28 | User information management method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110117719.3A CN113763019A (en) | 2021-01-28 | 2021-01-28 | User information management method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113763019A true CN113763019A (en) | 2021-12-07 |
Family
ID=78786497
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110117719.3A Pending CN113763019A (en) | 2021-01-28 | 2021-01-28 | User information management method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113763019A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023123933A1 (en) * | 2021-12-30 | 2023-07-06 | 深圳前海微众银行股份有限公司 | User type information determination method and device, and storage medium |
| CN116805253A (en) * | 2023-08-18 | 2023-09-26 | 腾讯科技（深圳）有限公司 | Intervention gain prediction method, device, storage medium and computer equipment |
| CN117035859A (en) * | 2023-08-14 | 2023-11-10 | 武汉利楚商务服务有限公司 | Intelligent releasing method and system for electronic coupons |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106022508A (en) * | 2016-05-06 | 2016-10-12 | 陈丛威 | Method and apparatus for predicting user friend invitation behaviors of online financing platform |
| CN109598410A (en) * | 2018-10-31 | 2019-04-09 | 平安科技（深圳）有限公司 | Presell methods of risk assessment, system, computer installation and readable storage medium storing program for executing |
| CN109993428A (en) * | 2019-03-28 | 2019-07-09 | 第四范式（北京）技术有限公司 | Resource allocation methods and device |
| WO2020020088A1 (en) * | 2018-07-23 | 2020-01-30 | 第四范式（北京）技术有限公司 | Neural network model training method and system, and prediction method and system |
| CN110766431A (en) * | 2018-07-27 | 2020-02-07 | 北京京东尚科信息技术有限公司 | Method and device for judging whether user is sensitive to coupon |
| CN111798273A (en) * | 2020-07-01 | 2020-10-20 | 中国建设银行股份有限公司 | Training method of purchase probability prediction model of product and purchase probability prediction method |
-
2021
- 2021-01-28 CN CN202110117719.3A patent/CN113763019A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106022508A (en) * | 2016-05-06 | 2016-10-12 | 陈丛威 | Method and apparatus for predicting user friend invitation behaviors of online financing platform |
| WO2020020088A1 (en) * | 2018-07-23 | 2020-01-30 | 第四范式（北京）技术有限公司 | Neural network model training method and system, and prediction method and system |
| CN110766431A (en) * | 2018-07-27 | 2020-02-07 | 北京京东尚科信息技术有限公司 | Method and device for judging whether user is sensitive to coupon |
| CN109598410A (en) * | 2018-10-31 | 2019-04-09 | 平安科技（深圳）有限公司 | Presell methods of risk assessment, system, computer installation and readable storage medium storing program for executing |
| CN109993428A (en) * | 2019-03-28 | 2019-07-09 | 第四范式（北京）技术有限公司 | Resource allocation methods and device |
| CN111798273A (en) * | 2020-07-01 | 2020-10-20 | 中国建设银行股份有限公司 | Training method of purchase probability prediction model of product and purchase probability prediction method |
Non-Patent Citations (1)
| Title |
|---|
| 胡晓丽;张会兵;董俊超;吴冬强;: "基于CNN-LSTM的用户购买行为预测模型", 计算机应用与软件, no. 06 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023123933A1 (en) * | 2021-12-30 | 2023-07-06 | 深圳前海微众银行股份有限公司 | User type information determination method and device, and storage medium |
| CN117035859A (en) * | 2023-08-14 | 2023-11-10 | 武汉利楚商务服务有限公司 | Intelligent releasing method and system for electronic coupons |
| CN116805253A (en) * | 2023-08-18 | 2023-09-26 | 腾讯科技（深圳）有限公司 | Intervention gain prediction method, device, storage medium and computer equipment |
| CN116805253B (en) * | 2023-08-18 | 2023-11-24 | 腾讯科技（深圳）有限公司 | Intervention gain prediction method, device, storage medium and computer equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR20200123015A (en) | Information recommendation method, apparatus, device and medium | |
| WO2022057658A1 (en) | Method and apparatus for training recommendation model, and computer device and storage medium | |
| CN113763019A (en) | User information management method and device | |
| CN110147882B (en) | Neural network model training method, crowd diffusion method, device and equipment | |
| CN110008973B (en) | Model training method, method and device for determining target user based on model | |
| WO2019072128A1 (en) | Object identification method and system therefor | |
| CN111798280B (en) | Multimedia information recommendation method, device and equipment and storage medium | |
| CN113065882A (en) | Commodity processing method and device and electronic equipment | |
| CN112036954A (en) | Item recommendation method and device, computer-readable storage medium and electronic device | |
| CN112446764A (en) | Game commodity recommendation method and device and electronic equipment | |
| CN114240555A (en) | Click rate prediction model training method and device and click rate prediction method and device | |
| CN111340522A (en) | Resource recommendation method, device, server and storage medium | |
| CN111047009A (en) | Event trigger probability pre-estimation model training method and event trigger probability pre-estimation method | |
| CN114493786A (en) | Information recommendation method and device | |
| CN109978594B (en) | Order processing method, device and medium | |
| CN112150184A (en) | Click rate estimation method and system, computer system and computer readable medium | |
| CN113743971A (en) | Data processing method and device | |
| CN113034168A (en) | Content item delivery method and device, computer equipment and storage medium | |
| CN111768218A (en) | Method and device for processing user interaction information | |
| WO2022156589A1 (en) | Method and device for determining live broadcast click rate | |
| CN113792952A (en) | Method and apparatus for generating a model | |
| CN113191806A (en) | Method and device for determining flow regulation target | |
| CN111325614B (en) | Recommendation method and device of electronic object and electronic equipment | |
| CN114663015A (en) | Replenishment method and device | |
| CN114049162A (en) | Model training method, demand prediction method, apparatus, device, and storage medium |
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 |