Information recommendation method and device

Abstract

The invention discloses an information recommendation method and device, and relates to the technical field of computers. A specific embodiment of the method comprises the steps of performing feature extraction on behavior data of a user in multiple fields to obtain user features, wherein the user features comprise a user flow feature, a user right behavior feature and a user ordering behavior feature; based on the user characteristics, obtaining the conversion probability of the user to each right; determining the order placing probability of the user based on the user flow characteristics and the user order placing behavior characteristics; and carrying out operation planning optimization calculation based on the conversion probability of the user to each right and interests and the order placing probability of the user to obtain to-be-recommended right and interests information of the user. According to the embodiment, the operation and maintenance cost is greatly reduced, and the flexibility of right and interest information recommendation is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method and device for information recommendation. Background Art

[0002] At present, the information recommendation of activities or rights (such as gift cards, coupons, points, etc.) for client users in the logistics field is mostly built with strategy combined with model as the basic framework. Among them, the strategy formed by business rules is the main framework, and the scoring of the model in different scenarios (such as order prediction, right usage prediction, etc.) is used as an auxiliary for comprehensive recommendation.

[0003] Although this idea and solution can effectively push rights and interests information to users to a certain extent, it requires users to configure policies and limits the flexibility of rights and interests information recommendations. Summary of the invention

[0004] In view of this, the embodiments of the present invention provide a method and device for information recommendation, which can apply a variety of dynamic data sources to construct features, greatly enriching user information; differentiated rights and interests information is recommended through an intelligent recommendation algorithm, and the recommendation and delivery of rights and interests information are automatically completed, without the need for operators to manually configure rights and interests information, greatly reducing operation and maintenance costs and improving the flexibility of rights and interests information recommendation.

[0005] To achieve the above object, according to one aspect of an embodiment of the present invention, a method for information recommendation is provided, comprising: Extracting features from the user's behavior data in multiple fields to obtain user features, including user traffic features, user rights and interests behavior features, and user ordering behavior features; Obtaining the conversion probability of the user to each benefit based on the user characteristics; Determine the order probability of the user based on the user traffic characteristics and the user order behavior characteristics; Based on the conversion probability of the user to each benefit and the order probability of the user, operations optimization calculation is performed to obtain the benefit information to be recommended for the user.

[0006] Optionally, obtaining the conversion probability of the user for each benefit based on the user characteristics includes: using a deep neural network based on a self-attention mechanism to process the user characteristics to obtain the conversion probability of the user for each benefit.

[0007] Optionally, the user characteristics also include user basic portrait characteristics; wherein, the user traffic characteristics are obtained by statistically analyzing the user's behavior data at different traffic ports; the user rights and interests behavior characteristics are obtained by statistically analyzing the user's rights and interests operation behavior data; the user ordering behavior characteristics are obtained by statistically analyzing the user's order quantity data; the user basic portrait characteristics are obtained by statistically analyzing the user's basic information and logistics frequency status data.

[0008] Optionally, determining the order probability of the user based on the user traffic characteristics and the user order behavior characteristics includes: inputting the user traffic characteristics and the user order behavior characteristics into a decision tree classification model to determine the order probability of the user.

[0009] Optionally, operations optimization calculations are performed based on the user's conversion probability for each benefit and the user's order probability to obtain the user's benefit information to be recommended, including: training an operations optimization model, and inputting the user's conversion probability for each benefit and the user's order probability into the operations optimization model to perform operations optimization calculations to obtain the user's benefit information to be recommended.

[0010] Optionally, the operations optimization model includes: decision variables: the decision result of recommending benefit j to user i; optimization objectives: maximizing the product of the user's order probability, the user's conversion probability to benefits, and the user's average standard income per order and the discount coefficient of the benefits; constraints include: each user can only choose one benefit, and the discount coefficient of the benefits issued to the user floats within a certain range.

[0011] According to another aspect of an embodiment of the present invention, there is provided an information recommendation device, comprising: A feature extraction module is used to extract features from the user's behavior data in multiple fields to obtain user features, wherein the user features include user traffic features, user rights and interests behavior features, and user ordering behavior features; A conversion prediction module, used to obtain the conversion probability of the user for each benefit based on the user characteristics; An order prediction module, used to determine the order probability of the user based on the user traffic characteristics and the user order behavior characteristics; The optimization calculation module is used to perform operations optimization calculation based on the conversion probability of the user to each benefit and the order probability of the user to obtain the benefit information to be recommended for the user.

[0012] According to another aspect of an embodiment of the present invention, there is provided an electronic device, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the information recommendation method provided by an embodiment of the present invention.

[0013] According to another aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for information recommendation provided by an embodiment of the present invention is implemented.

[0014] According to another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the information recommendation method provided by an embodiment of the present invention is implemented.

[0015] An embodiment of the above invention has the following advantages or beneficial effects: user characteristics are obtained by extracting features from user behavior data in multiple fields, and user characteristics include user traffic characteristics, user rights and interests behavior characteristics, and user order behavior characteristics; based on user characteristics, the user's conversion probability for each right and interest is obtained; based on user traffic characteristics and user order behavior characteristics, the user's order probability is determined; based on the user's conversion probability for each right and interest and the user's order probability, operations optimization calculations are performed to obtain a technical solution for the user's rights and interests information to be recommended, and a variety of dynamic data sources can be used for feature construction, greatly enriching user information; differentiated rights and interests information is recommended through an intelligent recommendation algorithm, and the recommendation and delivery of rights and interests information are automatically completed, without the need for operators to manually configure rights and interests information, greatly reducing operation and maintenance costs and improving the flexibility of rights and interests information recommendation.

[0016] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention. Figure 1 is a schematic diagram of main steps of a method for information recommendation according to an embodiment of the present invention; Figure 2 It is a schematic diagram of the implementation process of information recommendation according to an embodiment of the present invention; Figure 3 is a schematic diagram of main modules of an apparatus for information recommendation according to an embodiment of the present invention; Figure 4 is an exemplary system architecture diagram to which embodiments of the present invention may be applied; Figure 5It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0019] It should be noted that the collection, collection, updating, analysis, processing, use, transmission, storage and other aspects of user personal information involved in the technical solution disclosed in the present invention are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information, network security and national security.

[0020] The existing methods of recommending information based on the strategy combined model have the following defects: 1. Low flexibility: The solution that uses strategy as the main framework and model scoring results to recommend equity information does not have high flexibility in recommendation results. For example, the recommendation strategy that uses user stratification (user life cycle status) as the outermost architecture may only offer flash sale coupons to new users, lost users, and silent users, but ignores price-sensitive users and low-frequency active users, resulting in the loss of such customers. This solution that uses business strategy as the outermost architecture will limit the operation of the algorithm and affect the ability to recommend information; 2. High pressure on operators and low fault tolerance: The current recommendation strategy requires operators to manually configure all equity products in the marketing background, including but not limited to free coupons, paid coupons, new member activities, etc., involving more than ten kinds of products. The configuration and maintenance pressure on operators is high. Once misconfigured, it may lead to failure to obtain recommendation results, affecting the online recommendation effect, and the fault tolerance rate is low; 3. Unable to maximize revenue: The current model scoring is only for specific scenarios, such as whether to place an order in the next two weeks, whether to use coupons in the next two weeks, etc., and does not combine rights and interests information, user information, and order information for comprehensive recommendations. If the user will place an order, then the results of issuing 20% ​​off coupons and 10% off coupons are the same. If the 20% off coupon is issued at this time, the revenue of the logistics company will be reduced. Therefore, it is impossible to make rights and interests recommendations that maximize revenue after comprehensively considering all information.

[0021] Based on the above industry status, the present invention proposes the following rights and interests information recommendation scheme: 1. User-side features are formed based on user traffic characteristics (including but not limited to visit behaviors of different pages of mini programs and h5 (HTML5), exposure and click behaviors of different activities, etc.), basic portrait features (including basic information such as user's occupation, age group, price-sensitive tags, etc.), rights and interests behavior characteristics (statistical data of users receiving and using different rights and interests), and ordering behavior characteristics (statistical data of ordering behavior, package information, consignment information, etc.); system-side features are formed based on static information of the rights and interests themselves and the recent comprehensive use of the rights and interests; 2. Use DNN (Deep Neural Network) + self-attention to build a user equity prediction model. The introduction of the attention mechanism aims to learn the cross-information between different types and amounts of equity, as well as the cross-information between users and the corresponding positions of equity, and cooperate with DNN to perform multi-layer information abstraction and aggregation, predict the user's conversion of different equity, and obtain the user's conversion probability of different equity; 3. Use GBDT (Gradient Boosting Decision Tree) to build a basic order model, perform binary classification prediction on the user's future logistics order demand, and obtain the user's order probability.

[0022] Finally, the above data such as the user's order probability, the user's conversion probability to different rights and interests, the price and cost of the rights and interests are input into the operations optimization model for comprehensive solution to find the information recommendation results that maximize profits.

[0023] Figure 1 FIG. 1 is a schematic diagram of the main steps of the information recommendation method according to an embodiment of the present invention. Figure 1 As shown, the information recommendation method of the embodiment of the present invention mainly includes the following steps S101 to S104.

[0024] Step S101: extracting features from the user's behavior data in multiple fields to obtain user features, the user features including user traffic features, user equity behavior features, and user order behavior features. In an embodiment of the present invention, user behavior data in multiple fields may be selected for feature extraction to more comprehensively characterize the user's preference for certain equity products, wherein the equity products include but are not limited to points, free coupons, paid coupons, flash sales, etc.

[0025] In an embodiment of the present invention, the user characteristics may include user basic portrait characteristics in addition to user traffic characteristics, user rights and interests behavior characteristics and user ordering behavior characteristics; wherein the user traffic characteristics are obtained by statistically analyzing the user's behavior data at different traffic ports; the user rights and interests behavior characteristics are obtained by statistically analyzing the user's rights and interests operation behavior data; the user ordering behavior characteristics are obtained by statistically analyzing the user's order quantity data; and the user basic portrait characteristics are obtained by statistically analyzing the user's basic information and logistics frequency status data.

[0026] In one embodiment of the present invention, among the user characteristics in multiple fields involved, the first is the user traffic characteristics, including the user's exposure and click behavior data corresponding to different pages, different buttons, different activities, etc. at different traffic ports (for example, including: express applets, e-commerce platform applets, h5 landing pages, etc.), and by performing statistical analysis on the user's behavior data at different traffic ports (for example, statistical year-on-year and month-on-month change trends, etc.), it is possible to explore the transfer of user preferences from the user's traffic changes, so as to determine the preference for different equity activities through a model.

[0027] Secondly, the basic user portrait features, the present invention covers the static user information processed by the e-commerce platform side, including age, occupation, gender, etc. At the same time, the present invention also uses the express business side to process the user's logistics life cycle status (for example, including active, lost, silent, potential customers, etc.), and logistics frequency status (for example: the frequency of express delivery by users in the past month, and these frequencies are divided into low, medium, high and other different levels). Then, from the dimension of logistics life cycle status, users are divided into promotional activity target users (for example: users are in the state of loss, silence, waiting for recall, etc.), frequency increase target users (for example: new users, active status), new target users (for example: users are potential customers), and users are divided into low, medium, and high frequency users from the dimension of logistics frequency status, etc. In addition, the present invention also calculates the user's price sensitivity based on the distribution of users in the quantity-price relationship curve corresponding to price and sales volume, aiming to differentiate customer groups with different marketing sensitivity.

[0028] The third is the user rights behavior characteristics, which are based on the user's receipt, purchase, and use of rights in the logistics domain (for example, free coupons, paid coupon packages, family cards, points, etc.), as well as participation in and completion of activities, and statistics on year-on-year and month-on-month trends to explore the distribution and transfer of user rights and activity usage, and recommend rights that they are "more interested in" from the user's perspective.

[0029] Finally, we use the user’s order behavior characteristics to extract valid information from the user’s recent express order sequence. In addition, the waybill data contains consignment information, and a simple single-layer natural language processing model is used to extract key information from the consignment information and perform statistical aggregation.

[0030] After obtaining the user features in the above several fields, the above static and dynamic features can be spliced ​​together as input information of each model in subsequent processing. Among them, when performing feature splicing, for example, these features can be spliced ​​horizontally.

[0031] Step S102: Obtain the conversion probability of the user for each benefit based on the user features. In this step, the user features used are the user features obtained by combining the features of several fields extracted in step S101.

[0032] In an embodiment of the present invention, obtaining the conversion probability of the user for each benefit based on the user characteristics may specifically include: using a deep neural network based on a self-attention mechanism to process the user characteristics to obtain the conversion probability of the user for each benefit.

[0033] Among them, the deep neural network DNN (Deep Neural Network) is a machine learning model that uses multiple layers of neurons for feature extraction and learning. Through multiple levels of feature learning and combination, it can extract higher-dimensional abstract features from the original data, thereby improving the generalization ability of the model and further improving the model's prediction accuracy for unknown data. In addition, DNN can effectively process large-scale data through technologies such as distributed computing and GPU (graphics processing unit), and is applied to big data fields such as image, speech, and natural language processing. Its network structure can be designed and expanded according to the needs of the problem, and can realize models of different scales and complexities, with high flexibility.

[0034] Self-Attention is a common technique in deep learning, mainly used to process sequence data, such as text in natural language processing. The self-attention mechanism establishes relationships between each word by calculating the similarity between each word and other words, and calculates the vector representation of each word in a weighted manner based on these relationships. In the self-attention mechanism, the representation of each word consists of three parts: query vector, key vector, and value vector. For each word, the query vector and key vector can be obtained by linear transformation, and then the similarity between them is calculated by a similarity function (such as dot product or scaled dot product). Then, these similarities are normalized to obtain attention weights, and these weights are used to weighted sum the value vector to obtain the final vector representation. The advantage of the self-attention mechanism is that it can capture the relationship between any two positions in the sequence, so it performs well in sequence modeling tasks. It has been widely used in many natural language processing tasks, such as machine translation, text generation, and sentiment analysis.

[0035] The modeling idea of ​​the present invention is to integrate the information abstraction, generalization ability and cross-information capture ability of DNN with Self-attention, learn the competitive relationship between different rights, the similarity between users, and the user's inclination between different rights in the training process, and predict the user's conversion probability for different rights. In the specific implementation, the feature vector obtained after the user features are captured by Self-attention through cross-information can be directly input into DNN to extract higher-dimensional abstract features through multiple hidden layers, and then obtain the user's conversion probability for each right.

[0036] Step S103: Determine the order probability of the user based on the user traffic characteristics and the user order behavior characteristics.

[0037] According to an embodiment of the present invention, determining the order probability of the user based on the user traffic characteristics and the user order behavior characteristics may specifically include: inputting the user traffic characteristics and the user order behavior characteristics into a decision tree classification model to determine the order probability of the user.

[0038] The Gradient Boosting Decision Tree (GBDT) model is a very robust classification model that can flexibly process various types of data, including discrete user portraits, labels, etc. Based on the user's traffic and order characteristics, a user order model can be quickly built to accurately predict the user's order behavior in the future. By predicting the user's order behavior in the future, the user's order probability in the future can be further determined, so that it is convenient to combine the attribute information of each right (for example: the effective time of the right, etc.) to determine the activity rights to be recommended.

[0039] Step S104: performing operations optimization calculation based on the user's conversion probability of each benefit and the user's order probability to obtain the user's benefit information to be recommended.

[0040] According to one embodiment of the present invention, when performing operations optimization calculations based on the user's conversion probability for each benefit and the user's order probability to obtain the user's benefit information to be recommended, it can specifically include: training an operations optimization model, and inputting the user's conversion probability for each benefit and the user's order probability into the operations optimization model to perform operations optimization calculations to obtain the user's benefit information to be recommended.

[0041] The operations optimization model aims to make the best decision on the most suitable benefit for the user by combining the conversion probability of the user for each benefit and the order probability of the user obtained in the above two steps S102 and S103 with information such as the income brought by the user using different benefits and benefit discounts.

[0042] According to an embodiment of the present invention, the operations optimization model includes: decision variables, optimization objectives and constraints. Among them, the decision variable is the decision result of recommending benefit j to user i; the optimization objective is to maximize the product of the user's order probability, the user's conversion probability of benefits, the user's average standard income per order and the discount coefficient of the benefits; the constraints include: each user can only choose one benefit, and the discount coefficient of the benefits issued to the user floats within a certain range. In addition, other variables can also be included in the operations optimization model.

[0043] Specifically, the decision variables are: , where i is the user, j is the equity, and U ij is the decision result of recommending benefit j to user i, that is, whether to recommend benefit j to user i; Other variables: To predict the score of a user's choice of a certain benefit through DNN+Self-attention (deep neural network based on self-attention mechanism), The probability of a user placing an order predicted by GBDT, is the discount factor of equity j (e.g. 20% off, 10% off), is the average standard income of user i; Optimization goal: maximize the user's order probability * the user's conversion probability to a certain benefit * the average standard income per order * the benefit j discount coefficient, that is: ; Constraints: 1) Each user can only choose one benefit, namely: ; 2) The discount coefficient for the rights issued to users floats within a certain range, namely: .

[0044] By performing operations optimization calculations on the user's conversion probability for each benefit and the user's order probability, combined with information such as the income generated by the user's use of different benefits and benefit discounts, we can obtain the user's information on benefits to be recommended.

[0045] The technical solution of the present invention can be applied to the rights and interests recommendation of express delivery applets, application APPs, and h5 (HTML5) landing pages, and differentiated free and paid rights and interests can be recommended for different users, so as to achieve the marketing goals of promoting customer activity, increasing order frequency, recovering lost customers, and attracting new customers, thereby comprehensively increasing the number of users and express delivery revenue.

[0046] The technical solution of the present invention has the following advantages over the existing technical solutions: 1. The existing technology requires manual configuration by operators, and is not feasible and accurate. It can only allocate rights based on simple business rules. After introducing the technical solution of the present invention, the background marketing algorithm can calculate the most suitable rights for users and recommend them, greatly reducing operation and maintenance costs; 2. The intelligent algorithm of differentiated recommendation of the present invention can make benefits recommendations according to the situations of different users with the goal of maximizing profits, which can promote customer activity, increase order frequency, and attract new customers. After its implementation, the average number of orders per user increased by 12.4%, producing significant results.

[0047] Figure 2 FIG. 1 is a schematic diagram of an implementation flow of information recommendation according to an embodiment of the present invention. Figure 2As shown, in one embodiment of the present invention, the implementation process of the information recommendation is mainly as follows: first, feature processing is performed to obtain user basic portrait features, user traffic features, user rights and interests behavior features, and user ordering behavior features; then, the user features obtained by splicing the user basic portrait features, user traffic features, user rights and interests behavior features, and user ordering behavior features are input into the deep neural network model DNN+Self-attention based on the self-attention mechanism to obtain the user's conversion probability for each right; after splicing the user traffic features and the user ordering behavior features, they are input into the decision tree classification model GBDT to obtain the user's ordering probability. Finally, the user's conversion probability for each right and the user's ordering probability are input into the operations optimization model, and operations optimization calculations are performed to obtain the user's rights and interests information to be recommended. Among them, the operations optimization model mainly includes decision variables, optimization objectives, and constraints.

[0048] Figure 3 FIG. 1 is a schematic diagram of the main modules of the information recommendation device according to an embodiment of the present invention. Figure 3 As shown, the information recommendation device 300 of the embodiment of the present invention mainly includes a feature extraction module 301, a conversion prediction module 302, an order prediction module 303 and an optimization calculation module 304.

[0049] A feature extraction module 301 is used to extract features from user behavior data in multiple fields to obtain user features, wherein the user features include user traffic features, user rights and interests behavior features, and user ordering behavior features; A conversion prediction module 302, used to obtain the conversion probability of the user for each benefit based on the user characteristics; An order prediction module 303 is used to determine the order probability of the user based on the user traffic characteristics and the user order behavior characteristics; The optimization calculation module 304 is used to perform operations optimization calculation based on the user's conversion probability of each benefit and the user's order probability to obtain the user's benefit information to be recommended.

[0050] According to an embodiment of the present invention, the conversion prediction module 302 may also be used to: process the user features using a deep neural network based on a self-attention mechanism to obtain the conversion probability of the user for each benefit.

[0051] According to another embodiment of the present invention, the user characteristics also include user basic portrait characteristics; wherein, the user traffic characteristics are obtained by statistically analyzing the user's behavior data at different traffic ports; the user rights and interests behavior characteristics are obtained by statistically analyzing the user's rights and interests operation behavior data; the user ordering behavior characteristics are obtained by statistically analyzing the user's order quantity data; the user basic portrait characteristics are obtained by statistically analyzing the user's basic information and logistics frequency status data.

[0052] According to another embodiment of the present invention, the order prediction module 303 may also be used to: input the user traffic characteristics and the user order behavior characteristics into a decision tree classification model to determine the order probability of the user.

[0053] According to another embodiment of the present invention, the optimization calculation module 304 can also be used to: obtain the user's rights and interests information to be recommended by training an operations optimization model and inputting the user's conversion probability for each rights and interests and the user's order probability into the operations optimization model to perform operations optimization calculations.

[0054] According to another embodiment of the present invention, the operations optimization model includes: decision variables: the decision result of recommending benefit j to user i; optimization objectives: maximizing the product of the user's order probability, the user's conversion probability of benefits, and the user's average standard income per order and the discount coefficient of the benefits; constraints include: each user can only choose one benefit, and the discount coefficient of the benefits issued to the user floats within a certain range.

[0055] According to the technical solution of the embodiment of the present invention, user characteristics are obtained by extracting features from the user's behavioral data in multiple fields, and the user characteristics include user traffic characteristics, user rights and interests behavior characteristics, and user ordering behavior characteristics; the user's conversion probability for each right and interest is obtained based on the user characteristics; the user's ordering probability is determined based on the user traffic characteristics and the user's ordering behavior characteristics; operations optimization calculations are performed based on the user's conversion probability for each right and interest and the user's ordering probability, and a technical solution for obtaining the user's rights and interests information to be recommended can apply a variety of dynamic data sources for feature construction, greatly enriching user information; differentiated rights and interests information is recommended through an intelligent recommendation algorithm, and the recommendation and delivery of rights and interests information are automatically completed without the need for operators to manually configure the rights and interests information, greatly reducing operation and maintenance costs and improving the flexibility of rights and interests information recommendation.

[0056] Figure 4 An exemplary system architecture 400 is shown to which the method or apparatus for information recommendation according to an embodiment of the present invention may be applied.

[0057] like Figure 4As shown, system architecture 400 may include terminal devices 401, 402, 403, network 404 and server 405. Network 404 is used to provide a medium for communication links between terminal devices 401, 402, 403 and server 405. Network 404 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0058] Users can use terminal devices 401, 402, 403 to interact with server 405 through network 404 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 401, 402, 403, such as shopping applications, logistics and transportation applications, search applications, instant messaging tools, email clients, social platform software, etc. (only examples).

[0059] The terminal devices 401 , 402 , and 403 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0060] Server 405 may be a server that provides various services, such as a backend management server that provides support for shopping websites browsed by users using terminal devices 401, 402, and 403 (for example only). The backend management server may extract features of the user's behavior data in multiple fields from the received information recommendation request and other data to obtain user features, the user features including user traffic features, user rights and interests behavior features, and user order behavior features; obtain the user's conversion probability for each right and interest based on the user features; determine the user's order probability based on the user traffic features and the user's order behavior features; perform operations optimization calculations based on the user's conversion probability for each right and interest and the user's order probability to obtain the user's rights and interests information to be recommended, and feed back the processing results (such as the rights and interests information to be recommended - for example only) to the terminal device.

[0061] It should be noted that the information recommendation method provided in the embodiment of the present invention is generally executed by the server 405 , and accordingly, the information recommendation device is generally disposed in the server 405 .

[0062] It should be understood that Figure 4 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0063] Reference below Figure 5 , which shows a schematic diagram of the structure of a computer system 500 of a terminal device or a server suitable for implementing an embodiment of the present invention. Figure 5The terminal device or server shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0064] like Figure 5 As shown, the computer system 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage part 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the system 500 are also stored. The CPU 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0065] The following components are connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, a modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 510 as needed, so that a computer program read therefrom is installed into the storage section 508 as needed.

[0066] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 509, and / or installed from the removable medium 511. When the computer program is executed by the central processing unit (CPU) 501, the above-mentioned functions defined in the system of the present invention are executed.

[0067] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0068] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0069] The units or modules involved in the embodiments of the present invention may be implemented in software or hardware. The units or modules described may also be provided in a processor. For example, they may be described as follows: a processor includes a feature extraction module, a conversion prediction module, an order prediction module, and an optimization calculation module. The names of these units or modules do not, in some cases, constitute limitations on the units or modules themselves. For example, the conversion prediction module may also be described as "a module for obtaining the conversion probability of the user for each benefit based on the user characteristics."

[0070] As another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiment; or it may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs. When the above one or more programs are executed by a device, the device includes: extracting features from the user's behavior data in multiple fields to obtain user features, the user features including user traffic features, user rights and interests behavior features, and user order behavior features; obtaining the user's conversion probability for each right and interest based on the user features; determining the user's order probability based on the user traffic features and the user's order behavior features; performing operations optimization calculations based on the user's conversion probability for each right and interest and the user's order probability to obtain the user's rights and interests information to be recommended.

[0071] According to the technical solution of the embodiment of the present invention, user characteristics are obtained by extracting features from the user's behavioral data in multiple fields, and the user characteristics include user traffic characteristics, user rights and interests behavior characteristics, and user ordering behavior characteristics; the user's conversion probability for each right and interest is obtained based on the user characteristics; the user's ordering probability is determined based on the user traffic characteristics and the user's ordering behavior characteristics; operations optimization calculations are performed based on the user's conversion probability for each right and interest and the user's ordering probability, and a technical solution for obtaining the user's rights and interests information to be recommended can apply a variety of dynamic data sources for feature construction, greatly enriching user information; differentiated rights and interests information is recommended through an intelligent recommendation algorithm, and the recommendation and delivery of rights and interests information are automatically completed without the need for operators to manually configure the rights and interests information, greatly reducing operation and maintenance costs and improving the flexibility of rights and interests information recommendation.

[0072] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for information recommendation, characterized in that: include: Extracting features from the user's behavior data in multiple fields to obtain user features, including user traffic features, user rights and interests behavior features, and user ordering behavior features; Obtaining the conversion probability of the user to each benefit based on the user characteristics; Determine the order probability of the user based on the user traffic characteristics and the user order behavior characteristics; Based on the conversion probability of the user to each benefit and the order probability of the user, operations optimization calculation is performed to obtain the benefit information to be recommended for the user.

2. The method according to claim 1, characterized in that Obtaining the conversion probability of the user for each benefit based on the user characteristics includes: The user characteristics are processed using a deep neural network based on a self-attention mechanism to obtain the conversion probability of the user for each benefit.

3. The method according to claim 1, characterized in that The user characteristics also include user basic portrait characteristics; wherein, The user traffic characteristics are obtained by statistically analyzing the behavior data of users at different traffic ports; The user rights and interests behavior characteristics are obtained by statistically analyzing the user's rights and interests operation behavior data; The user ordering behavior characteristics are obtained by statistically analyzing the user's order volume data; The user basic portrait features are obtained by statistically analyzing the user's basic information and logistics frequency status data.

4. The method according to claim 1, characterized in that: Determining the order probability of the user based on the user traffic characteristics and the user order behavior characteristics includes: The user traffic characteristics and the user ordering behavior characteristics are input into a decision tree classification model to determine the ordering probability of the user.

5. The method according to claim 1, characterized in that Based on the conversion probability of the user to each benefit and the order probability of the user, operations optimization calculation is performed to obtain the benefit information to be recommended for the user, including: By training an operations optimization model, and inputting the user's conversion probability for each benefit and the user's order probability into the operations optimization model to perform operations optimization calculations, the user's benefit information to be recommended is obtained.

6. The method according to claim 5, characterized in that The operations research optimization model includes: Decision variable: the decision result of recommending benefit j to user i; Optimization goal: maximize the product of the user's order probability, the user's conversion probability to equity, the user's average standard income per order, and the discount coefficient of the equity; The constraints include: each user can only choose one benefit, and the discount coefficient of the benefits issued to users fluctuates within a certain range.

7. An information recommendation device, characterized in that: include: A feature extraction module is used to extract features from the user's behavior data in multiple fields to obtain user features, wherein the user features include user traffic features, user rights and interests behavior features, and user ordering behavior features; A conversion prediction module, used to obtain the conversion probability of the user for each benefit based on the user characteristics; An order prediction module, used to determine the order probability of the user based on the user traffic characteristics and the user order behavior characteristics; The optimization calculation module is used to perform operations optimization calculation based on the conversion probability of the user to each benefit and the order probability of the user to obtain the benefit information to be recommended for the user.

8. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

9. A computer readable medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images